managing human resources in a decentr alized context

A central rationale for decentralization is that,by moving government closer to the people, itbrings public sector activities and decisions in linewith popular preferences. Yet, in practice, the civilservice—a critical component of government—rarely enters the decision calculus of decentraliza-tion design. The civil service is often recognized asan afterthought rather than seen as essential to suc-cessful decentralization. This is lamentable but notsurprising, as decentralization is quintessentially apolitical process. In East Asia, whether it was thedesire to quell the forces of regional disintegrationin the Philippines and Indonesia, the urgent needto meet the demands of economic transition inChina and Vietnam, or pressure to improve servicedelivery and citizen participation in Cambodia andThailand, the primary motivation for decentraliza-tion has been political—notwithstanding variedproximate causes.

The common failure to address the details ofcivil service management as an integral part of thedecentralization package has significant implica-tions. Civil servants form a crucial link between thedelivery of financial resources to the governmentand the delivery of essential public services tothe people (World Bank 2003a). The relation-ship between decentralization and civil service

129

7

Managing HumanResources in aDecentralized

ContextAmanda E. Green

management is a two-way process. The behavior ofcivil servants has important consequences for gov-ernment performance in a decentralized setting.Conversely, decentralization alters both the incen-tives of and the demands on the civil service.Managing this behavior is critical to realizing thebenefits of decentralization.

Accordingly, this chapter argues that civilservice management—or more broadly humanresource management—should be seen as anessential component in the design of decentraliza-tion rather than a separate, stand-alone process.The discussion marshals evidence from across EastAsia and around the world. The following sectionpresents a framework for exploring the interactionbetween administrative decentralization and civilservice management—in terms of both the theo-retical goals of decentralization and their realiza-tion in practice. The third section delves into therealities of administrative decentralization in EastAsia through the experiences of six countries:Cambodia, China, Indonesia, the Philippines,Thailand, and Vietnam. The varied approachesand responses of these countries offer insightsinto the process of managing civil servants in adecentralized setting. The final section examinesthe central dilemmas that arise in designing

east_129-154.qxd 5/17/05 11:42 AM Page 129

130 East Asia Decentralizes

administrative decentralization and draws lessonson how countries in East Asia and beyond canmaximize the benefits and minimize the risks ofdecentralization.

The Interplay betweenDecentralization and HumanResource Management

Decentralization is a spectrum rather than a singlestate, ranging from deconcentration to delegationto devolution (see box 1.1). As table 7.1 shows,key features of the civil service vary with thedegree of administrative decentralization. Overall,East Asian countries fall into the intermediate cat-egory, with local managers enjoying some freedomto recruit and allocate staff subject to centralguidelines on pay levels and total employment.Indonesia and the Philippines took the biggestleaps along the decentralization continuum inboth law and practice, but central governmentsretained considerable control over civil servicewages at the local level. In Indonesia, recent revi-sions to the Law on Regional Administration haverolled back decentralization in hiring decisions.In the Philippines, the momentum behind imple-

menting decentralization policy has slowed con-siderably. China and Vietnam have taken a moreopportunistic approach, experimenting with dif-ferent degrees of local autonomy. However, thecentral government and Communist Party con-tinue to influence how decentralization plays out.In Thailand, a legal framework defines an interme-diate form of administrative decentralization, butmost of this framework has yet to materialize inpractice. Finally, Cambodia can best be describedas a deconcentrated system, with a high propor-tion of staff based in the field but working onbehalf of the central government.

Movement along the spectrum of administrativedecentralization depends partly on interplay withthe political and fiscal aspects of decentralization.For example, a local government that has fullauthority over the size of its civil service cannonetheless face restrictions in the use of fundstransferred from the center. Similarly, the strengthof a country’s accountability framework is influ-enced by whether political decentralization has cre-ated institutions for locally elected politicians tooversee the activities of local governments. The casestudies illustrate the results of a mismatch in thedimensions of decentralization.

TABLE 7.1 Key Features of Administrative Decentralization

Deconcentration (minimal change)

• Provider staff working at locallevel are employees of andaccountable to the center,usually through theirrespective ministries; centralemployees compensate forweak local capacity.

• Accountability remainsdistant: the short route ofaccountability may be weak ifprovider monitoring is weak,and citizens may have torely on a weak long routestretching to politicians at thecenter; a strong compactbetween policymakers andproviders can compensate tosome extent.

Delegation (intermediate change)

• Providers could be employeesof central or local government,but the center typically definespay and employment.

• Local government has someauthority over hiring andlocation of staff, but is lesslikely to have authority overfiring.

• Both long and short routes ofaccountability are potentiallystronger; greater localknowledge can allow bettermatching of supply with localpreferences and bettermonitoring, strengtheningboth the compact and clientpower.

Devolution(substantial change)

• Providers are employees oflocal government.

• Local government has fulldiscretion over salary levels,allocation, and numbers ofstaff, as well as the authorityto hire and fire.

• An overarching civil serviceframework covering localgovernments may stillestablish standards andprocedures for hiring andmanaging staff.

• Potentially strongest long andshort routes to accountability,but influenced by local socialnorms and vulnerable to localcapacity constraints andpolitics.

Source: Adapted from World Bank 2003h, 189.

Administrative Decentralization

east_129-154.qxd 5/17/05 11:42 AM Page 130

Managing Human Resources in a Decentralized Context 131

What Makes a Decentralized Civil Service Work?1

To better understand the opportunities and obsta-cles that a country may encounter on the road toadministrative decentralization, it is first useful toconsider the destination. A functioning system ofdecentralized civil service management—as dis-tinct from a centrally directed model—has severalkey characteristics:

• Local government functions are clearly definedso staff know what is expected of them and man-agers can adapt the local civil service to reflectwhat needs to be done, without inefficient gapsor overlap with other levels of government.

• Local government can allocate staff across func-tions as needed. This requires that civil servicemanagers have autonomy—or at least influ-ence—in operating the local establishment aswell as deploying staff across different depart-ments and facilities.

• Local government is able to attract and retainqualified individuals, and to build a team with adiverse set of skills. This requires that localgovernment have something to offer, throughcompetitive pay, career opportunities, prestige,or other incentives.

• Local government has flexibility in managingfinancial resources. Managing civil servantsrequires managing their cost, either directlythrough pay levels or indirectly through staffnumbers.

• Local government can hold staff accountable fortheir performance. This requires the capacityto supervise and monitor civil servants, the abil-ity to reward good performance through payincreases, promotions, and other benefits, andthe authority to punish deficient performancethrough disciplinary measures or dismissals.

Countries may not meet some or all of thesecriteria for numerous reasons. In some cases, thefailure is one of design. For example, as mentioned,comparatively limited independence among localgovernments on fiscal and political matters mayconstrain their administrative autonomy.

In other cases, it is by design that the criteria arenot met. First, keeping local civil services underthe umbrella of central direction can ensure somestandardization in working conditions across the

country. Otherwise, local governments in poorerareas will find it difficult to compete against richerareas, and national coverage of public services maysuffer. Second, central government involvement instaffing may expand civil servants’ career paths byopening channels between local and centralemployment. Third, the center may wish to retaincontrol over hiring and pay to shield subnationalgovernments from local pressure to overspend onwages and salaries. Fourth, centralized rules can beimportant in sectors that need to enforce minimumprofessional qualifications, such as with teachers,doctors, and nurses. Fifth, where ethnic or othertensions threaten stability, the central governmentmay use the civil service as a tool for nationalintegration. Finally, the central government’s reluc-tance to decentralize may stem simply from a desireto retain control.

The Implications of Decentralization

How a country resolves the tension between themotivations for decentralizing the civil service onthe one hand, and the reasons for caution on theother, will shape the design of decentralization and,ultimately, how the process plays out. The implica-tions of decentralizing civil service managementcan be examined along four critical dimensions:capacity, incentives, autonomy, and accountability.Each factor plays a significant role in the success ofdecentralization and is, in turn, heavily influencedby the decentralization process.

These four dimensions are closely linked andentail important tradeoffs. For example, civil serv-ice training programs are not likely to strengthencapacity in a sustainable way unless incentives moti-vate civil servants to use what they learn. Similarly,improvements in accountability at the local levelrequire that civil servants have the capacity—through accounts and records—to render thataccountability effectively. Finally, holding local civilservice managers accountable for their decisions isdifficult when they do not have autonomy in mak-ing those decisions.

Capacity. For civil servants to deliver the higher-quality local services envisioned underdecentralization, they need to have the capacity todo so. This involves both individual and institu-tional elements. First, the success of decentraliza-tion depends on the ability of individual civil

east_129-154.qxd 5/17/05 11:42 AM Page 131

servants to take on new tasks, at both central andlocal levels. Second, smaller size and budgets canconstrain the institutional capacity of local govern-ments. The process of decentralization itself canhave important implications for capacity needs atthe local level. The devolution of public serviceresponsibilities requires both a broader variety ofskills and a greater depth of knowledge in specificareas, such as financial management and perfor-mance monitoring. Moreover, local leaders need tolearn to supervise staff, mobilize more own-sourcerevenues, interact with local constituents andelected officials, and develop local institutions. For-merly central civil servants who move to local levelscan transfer some of these skills; local governmentsneed to gain the rest through experience. At thesame time, central employees need to shift from“doing” to facilitating and supervising.

Incentives. In some cases, what appears to be alack of capacity to carry out the functions of decen-tralized government is instead a lack of motivationto act in the public interest. The structure and man-agement of the civil service influence the outcomeof decentralization reforms by affecting how localcivil servants behave. The level of pay and benefits,options for career mobility, and degree to whichmerit is recognized or unsatisfactory performancepenalized can determine the dedication withwhich a civil servant works, as well as the type ofindividuals who choose to become civil servants.Decentralizing functional and managementresponsibilities to the local level, in turn, modifiesthe incentive structures of local civil servants. Theproximity of local government to the recipients ofpublic services can tighten the link between effortsand results. However, some local civil services aretoo small to offer significant opportunities forcareer advancement, and poorer local governmentsmay be unable to pay salaries high enough toattract talent. In remote areas, a combination oflow pay and difficult conditions creates a viciouscircle whereby an inability to attract high-qualitystaff leads to further deterioration in conditions.

Autonomy. The argument that decentralizationboosts the responsiveness of civil servants assumesthat local managers have the authority to respond tothe demands of their constituents. Local autonomyin allocating human resources can improve effi-ciency by allowing managers to hire staff whoseskills align with planned activities, to discipline or

dismiss ineffective staff, and to trim numbers tokeep costs down. Though less common at the locallevel, direct financial autonomy—such as the abilityto set pay levels and charge user fees—can improvestaff performance and thereby enhance the benefitsof decentralization. However, to achieve these bene-fits, performance incentives and accountabilityframeworks must be robust enough to preventinefficiency and mismanagement. By definition,administrative decentralization would seem to aug-ment local autonomy, but this is not always the case.The center often retains a significant degree ofcontrol, particularly in financially sensitive areassuch as the size and wage levels of the civil service,especially when local capacity and accountability arein question. Table 7.2 provides a stylized but usefulcomparison of East Asian countries in these areas.

Accountability. The potential for improvingservice delivery through decentralization depends,among other factors, on accountability relation-ships—the degree to which civil servants are heldto account for their performance and integrity, andto whom they are accountable. Without a strongsystem of local accountability, devolving authorityand financial resources to local governments canlead to waste or misuse of public funds, and thepotential for political capture at the local level candistort the benefits of decentralization. On theother hand, where corruption is systemic at thecentral level, devolution may enhance service deliv-ery. Where political decentralization allows foroversight by locally elected bodies, the need forreelection offers strong incentives for better per-formance. If civil society monitoring mechanismsare strong, the downward accountability of localstaff will tend to encourage a closer connectionbetween public services and citizen demands. Ifaccountability is only to the center, however, decen-tralization may not deliver the potential benefits ofbringing government closer to the people.

Decentralization can, in turn, affect localaccountability. In shifting responsibility for moni-toring civil servants to local managers, decentral-ization can make it more difficult for civil servantsto get away with laziness and corruption. Thecloser proximity of local citizens to governmentdecision makers can enhance their ability to holdthose decision makers accountable. However, ifcritical checks and balances to protect the neutral-ity and independence of civil servants are not in


east_129-154.qxd 5/17/05 11:42 AM Page 132

place, decentralization can lead to nepotism, withlocal managers rewarding family members andsupporters with coveted positions, and can facili-tate political capture by bringing civil servantswithin reach of local power bases.

Taken together, the capacity, incentives, auton-omy, and accountability of civil service manage-ment provide both a lens through which to evaluatethe design of decentralization and a picture of howadministrative decentralization has played out inpractice. The following section shows how thesefour dimensions have influenced decentralizationin East Asian countries, and how policy makers canleverage those dimensions to get the most out ofdecentralization.

Civil Service Managementon the Ground in East Asia

The structure of government and, by extension, thecivil service varies greatly across East Asia. Somecountries have focused decentralization efforts onthe lowest levels of government, while others haveemphasized deconcentration to an intermediatelevel, such as provinces. Countries also differ in

whom they consider civil servants. Some countriestreat teachers, health workers, and police as part ofthe civil service, while others consider them separatecategories. The distinction between ministry staffon the one hand, and employees of public serviceagencies and state-owned enterprises on the other,may also blur the definition of a civil servant. Finally,the determination of which civil servants are subna-tional employees can reflect either their physicallocation or the level of government that pays them.

Even in the absence of agreed standards anddefinitions, broadly characterizing civil servicedecentralization in the region is worthwhile.Figure 7.1 shows the wide variation in civil servicestructures across the region. The share of employ-ees at the subnational level ranges from around19 percent in Thailand to just over 90 percentin China. The share of personnel expenditures intotal subnational spending also reflects the extentof administrative decentralization. As figure 7.2shows, average subnational spending on personnelranges from roughly 41 percent of total spendingin Thailand to around 60 percent in China. It isimportant to note, however, that these figures donot necessarily point to China as the region’s most


TABLE 7.2 Central Authority over the Subnational Civil Service

Cambodia China Indonesia Philippines Thailand Vietnam

Legislation and regulations:

Designation 1 2 1 2–3 2 1Recruitment 1 4 2 3 2 2–3

Structure and career management:

Establishment control 1 3 2 2–3 2 1Appointment and

mobility 2 3 3 4 2 3Employment

framework 1 1 1 3 1 1

Performance management:

Standard setting and rewards 2 2 2 3 2 2

Training and development 2 3 3 4 3 3

Accountability 2 4 3 4 2 2

Sources: World Bank 2003d, 16; staff estimates.Note: 1� total central authority; 2 � central dominance; 3 � central guidance; 4 � central leadership;5 � autonomous.

east_129-154.qxd 5/17/05 11:42 AM Page 133

and innovative responses, the cases attempt to gleanimportant lessons about the interplay betweencapacity, incentives, autonomy, and accountabilityin decentralized civil service management.

The Big Bang and Beyond: The Philippinesand Indonesia

In both the Philippines and Indonesia, decentral-ization was an integral part of a political open-ing following the overthrow of an authoritarianregime. In these sprawling archipelagos, risingsocial tensions added a note of urgency to the deci-sion to decentralize, and local autonomy came to beseen as the key to quelling threats to national unity.Neither government felt it had the luxury of per-fecting the design of intergovernmental fiscal andadministrative arrangements before decentralizing.Following the logic that a chasm cannot becrossed in two leaps has worked well in manyways.2 Neither country experienced major servicedisruptions during the transition, and the rela-tively open-ended design of decentralization hasencouraged innovative local responses to newresponsibilities. However, once the dust settled,momentum slowed, and important challengesremain in correcting some of the imperfections


decentralized country. The degree of authorityexercised by subnational civil service managers indetermining the amount, structure, and allocationof resources plays an important role. Though dif-ferent countries have focused their decentralizationpolicies on different levels of government, table 7.3provides a general picture of how civil service man-agement practices have taken shape in East Asia.

The remainder of this section outlines the expe-riences of East Asian countries in dealing with thehuman resource implications of decentralization. InIndonesia and the Philippines, extensive decentral-ization of the administrative apparatus has broughtto light both the opportunities and the risks ofdevolving authority to local managers. In Chinaand Vietnam, where decentralization has proceededin stages, the roles of state and party offer additionalinsights into the relationship between decentraliza-tion and civil service management. Finally, in Cam-bodia and Thailand, the focus on building capacitybefore rather than through decentralization hasmeant limited implementation of decentralizationpolicies, and this experience offers an opportunityto consider how best to balance caution andprogress. Together these six case studies explore therealities of human resource management in adecentralized setting. In addressing both pitfalls

FIGURE 7.1 Subnational Employees as aShare of Total Government Employment

per

cent

0

10

20

40

60

80

100

30

50

70

90

China

Indo

nesia

Vietn

am

Cambo

dia

Thail

and

Philip

pines

Sources: Cambodia—World Bank 2003a (includes politicalappointees and nonpermanent staff); China—World Bank 2002a,annex 2 (includes Public Service Units outside health and education); Indonesia—World Bank 2003d; Philippines—WorldBank 2003f; Thailand—Kingdom of Thailand, Office of the CivilService Commission; Vietnam—Government of Vietnam, GeneralStatistical Office, 2003.Note: Data are for the most recent available year, ranging from2000–3, and exclude staff in the education, health, military, andpolice sectors, unless otherwise noted.

FIGURE 7.2 Personnel Spending as a Shareof Total Subnational GovernmentExpenditures

per

cent

0

10

20

40

60

30

50

70

China

Cambo

dia

Philip

pines

Thail

and

Indo

nesia

Sources: Cambodia—World Bank 2003a; China—World Bank2003b; Indonesia—World Bank 2003d; Philippines—World Bank2003f; Thailand—Weist 2003.Note: Data are for the most recent available year, ranging from2000–3.

east_129-154.qxd 5/17/05 11:42 AM Page 134

brought to light by the decentralization process.As East Asia’s two most rapid decentralizers, thePhilippines and Indonesia provide a useful startingpoint for addressing administrative decentraliza-tion in the region.

Central Rules and Local Responses in thePhilippines. The 1991 Local Government Code(LGC) provided the major impetus for“genuine andmeaningful local autonomy” in the Philippines,


TABLE 7.3 Staffing Authority among Subnational Governments

Enabling mechanisms Cambodia China Indonesia Philippines Thailand Vietnam

Budget control• Determine the wage

envelope � � � � � �

• Dismiss surplus staff � � � � � �

Establishment control• Control overall

staffing numbers � � � � � �

• Control staffing numbers in individual offices and facilities � � � � � �

Recruitment• Formal employer � � � � � �

• Have authority to hire � � � � � �

• Have independent merit-based recruitment mechanism (local civil service commission) � � � � � �

Career management• Promotion is available � � � � � �

• Transfers within local government are possible � � � � � �

• Horizontal mobility within local civil service cadre � � � � � �

Performance management• Direct and supervise

activities and tasks � � � � � �

• Conduct evaluations � � � � � �

• Offer financial rewards � � � � � �

• Discipline and fire underperforming staff � � � � � �

Pay policy• Set overall wage rates � � � � � �

• Set local incentives/salary top-ups � � � � � �

Source: Author’s estimates.Note: � � yes; � � partial; � � no. Data are for most recent available year, ranging from 2000 to 2003.Ratings refer to the subnational level prioritized by the country’s decentralization policy and to de factopractices as well as de jure authority.

devolving substantial government functions, andattendant financial and human resources, from thenational to the local level.3 Over 70,000 staff weretransferred to local governments.

In each tier of government, the local chiefexecutive—governor, mayor, or barangay captain—can hire, fire, and promote staff, subject to guide-lines from the central Civil Service Commission(CSC).4 Civil servants are bound to a code of con-duct and required to declare their net worth and

east_129-154.qxd 5/17/05 11:42 AM Page 135

financial and business interests. Governmentemployees are also supposed to disclose a list ofclose relatives in the public service, but this doesnot in practice seem to limit family connections inthe civil service. Indeed, despite the constitutionalrequirement that “appointments in the civil serviceshall be made according to merit and fitness,”several mechanisms cause merit and fitness to loseout to patronage and nepotism at the local level.5

For example, although candidates on the shortlistsfor civil service appointments generally meetthe minimum qualifications set out by the CSC, thelocal chief executive is not bound to choose thetop-ranked applicant.

Moreover, recruitment for confidential, highlytechnical, and noncareer staff positions is exemptfrom the prescribed appointment process. Perhapsbecause these workers are relatively easy to hire,noncareer employees composed over one-third(38.6 percent) of total local government staff in2001—significantly more than in national govern-ment agencies (4.67 percent) (World Bank 2003e).Furthermore, local governments can retain emer-gency and casual staff for up to six months withoutCSC approval. Local officials often use this provisionto circumvent central controls and delays andavoid contributing to the national employee benefitscheme. Repeated extensions of temporary contractslead to de facto permanent employment, and not alllocal governments advertise nonpermanent posi-tions. This lack of transparency undermines merit-based hiring, as temporary positions are oftenhanded out as patronage for loyal supporters.

Central regulations on local establishment sizeand expenditure allocations constrain the chiefexecutive’s flexibility in appointing staff. A uniformsystem of job classification imposes rigidities onsmall local civil services. The central governmentlimits local personal services expenditures to 45 to55 percent of the previous year’s income, depend-ing on the income class of the local government.6

The center also determines salary scales and benefitentitlements. Base wages tend to be low, but aresupplemented by a complex system of both mone-tary and in-kind allowances and benefits. From1992 to 2001, aggregate local spending on personalservices averaged 56.8 percent of the previous year’stotal regular income, exceeding the nationally man-dated cap on the wage bill. In municipalities, theaverage was 64.4 percent (World Bank 2003f).7

The 1989 Compensation and Position Classifi-cation Act set salaries in all but first-class local gov-ernments lower than those at the center.8 However,because the salaries of national staff transferred tolocal governments remained the same, this createda wage gap between local and devolved staff. Infact, some civil servants who transferred to lower-income municipalities earned more than themayor. This influx of better-paid staff imposeda budget crunch on local governments. The 1993Salary Standardization Law, which called for unify-ing pay regimes across all levels of government,exacerbated this problem by raising salaries atlower levels. The law has also affected the incentivesof civil servants by compressing salaries and thuslowering the increases that employees can expect asthey move up the ranks. The compact organiza-tional structure of local governments also restrictsthe career mobility of civil servants.

Local governments have responded in differentways to these central directives. Some cash-strappedgovernments are forced to simply ignore the man-dated salary scales and pay their employees less.Some local chief executives have elected not to fillmandatory positions to leave room for other staff orsalary supplements. Others have attempted to layoff workers, though this is rare owing to the politi-cal sensitivity of retrenchments. Local governmentswith greater access to own-source revenues supple-ment staff incomes even for positions covered bythe national government, such as police. Local gov-ernments have also responded to rising personnelcosts by charging some expenditures, such as pay-ments for contract workers, to other budget lines.This crowds out nonpersonnel expenditures whilemasking the true extent of staff costs. As the centralgovernment does not have the capacity to enforceits regulations, local governments have much moreflexibility than the legal framework implies.

Balancing National Unity and Local Autonomy inIndonesia. Indonesia’s Big Bang decentralization—ambitious and swift as it was—risked disruptingpublic services, inciting local unrest, and leavingcivil servants unpaid. In actuality, the transitionwent quite smoothly. Delivery of services contin-ued for the most part unscathed, and expectedupheavals among reassigned civil servants failed tomaterialize. Over 2.1 million staff—nearly 60 per-cent of central employees—were transferred to


east_129-154.qxd 5/17/05 11:42 AM Page 136

provincial and (primarily) district civil services withrelatively little difficulty, although few people had tophysically move because the previous system hadbeen highly deconcentrated. The central govern-ment greatly aided the process by continuing to paydevolved civil servants from national coffers for fivemonths.

In the wake of the Big Bang, several legal dis-crepancies came to light. Considerable confusionensued in applying various laws and regulations,leading to overlapping authority and eroding effi-ciency. According to the original administrativeframework (Law 22 of 1999 on Regional Adminis-tration), the head of region had the right to hire,fire, promote, transfer, and discipline staff. TheCivil Service Law, however, maintained thatauthority for civil service appointments rested firstand foremost with the central government, whichcould delegate that authority to heads of region.Government Regulation 97 of 2000 allowed thehead of the regional administration to determinethe size of the regional establishment. However,central regulations requiring nationwide advertis-ing for some posts—although perhaps intended tospark interregional mobility—limited the flexibil-ity of regional managers.

Recent changes to the framework laws havemoderated the initial thrust toward devolution andintroduced even more uncertainty about the rolesof different levels of government in managing thecivil service. The revised Law on Regional Adminis-tration (Law 32 of 2004) has rolled back localauthority to manage human resources, stipulatingthat the central government will supervise regionalcivil servants within a national framework, and thatregional heads can hire, fire, and relocate onlylower-level staff at the same level of government.9

Regional managers no longer have explicit author-ity to manage their civil services.

The central government also determines paylevels and increases regardless of their affordabilityat the local level, limiting the autonomy of regionalmanagers.10 Local governments must further paythe pensions of civil servants transferred from thecenter, which can create an unsustainable burden,particularly as the cost of pensions rises along withsalaries. In 2001, the dana alokasi umum (DAU)—or general allocation grant—was supplemented bya “contingency fund” to cover the 14–30 percentpay increase mandated by the center. Though the

central government introduced this transitionalmechanism with good intentions, it created incen-tives for overstaffing at the local level, as regionalgovernments began to assume that excess person-nel costs would continue to be covered by supple-mentary funding.

The structure of the DAU itself discourages pru-dent local management. Currently, the DAU isdivided into three parts: a “lump sum” distributedin equal amounts to every region, a “formula”amount determined by the extent to which aregion’s financial need outstrips its fiscal capacity,and a “balancing factor” based largely on theregion’s wage bill as a share of total subnationalwage expenditures in the previous year. The balanc-ing factor deters local managers from cutting costsbecause lowering the wage bill lowers next year’sDAU allocation. This disincentive will becomestronger with the implementation of the revisedLaw on Fiscal Balance (Law 33 of 2004). The equal-izing lump sum component will be eliminated in2006, and the “hold harmless” clause that keepsregional DAU allocations from decreasing overtime applies only until 2008. Moreover, in the inter-est of national unity, the central government hasresolved to fully fund regional salaries through theDAU, which will create strong incentives for subna-tional managers to increase staffing.

Until recently, a zero-growth policy preventedformal recruitment to the Indonesian civil service.New staff appointments in November 2004 werethe first of their kind since decentralization wasimplemented. At the same time, Government Reg-ulation 8 of 2003 on Local Government Structureand Organization defines the maximum number ofstaff in the regions, though it is not clear howstrictly this regulation is enforced. Both the centraland regional governments have taken to hiringcontract workers to circumvent these formalrestrictions. At the local level, this strategy hasallowed managers to hire and fire temporary staffand consultants as desired, without interferencefrom the center. Contract workers, who nowaccount for about 10 percent of governmentemployees, are also attractive because their employ-ment is not subject to qualification standards orpension obligations.

In theory, the government must base hiring andpromotion decisions on merit, but anecdotalevidence reveals that this needs strengthening in


east_129-154.qxd 5/17/05 11:42 AM Page 137

practice, especially in promoting staff. The head ofregion has the right to discipline and dismiss lower-level staff who are underperforming, subject to dueprocess and approval from the provincial governor.Dismissals are rare in practice, however, and staffreassignment seems to be the preferred penalty.

The civil service agency, BKN, has begun todefine job classifications and qualifications for thecivil service as a whole, including regional govern-ments. The Indonesian government has also beendeveloping minimum standards for service deliv-ery, including broad directives on supervising min-imum wage levels as well as highly specific technicalguidelines. Heads of region have generally beenenthusiastic about this initiative, both because theyprefer to be held accountable against establishedcriteria rather than arbitrary standards set by localparliaments, and because they expect to receivemore funding in exchange for taking on greaterresponsibilities (World Bank 2003d). In principle,minimum service standards can improve civil serv-ice management by offering some basis for unityamong public servants, ensuring a minimum levelof performance across the country, and enhancinglocal accountability through improved understand-ing of what is expected of local civil servants. InIndonesia, however, most line ministries are issuingstandards without giving enough thought to theirfeasibility and affordability at the local level. To beeffective, minimum standards must be designedcarefully, so they are specific enough to provideclear direction on what is expected of the local civilservice but not so detailed as to hamstring regionalleaders.

The design of decentralization in Indonesia con-stricts the interregional mobility of civil servants,and therefore their career paths. No streamlinedsystem exists for relocating subnational civil ser-vants to other areas. Discriminatory hiring prac-tices at the regional level seem to further limitmobility. Despite regulations designed to preventthis, some heads of region are reportedly engagingin preferential treatment of “sons of the soil,” ordiscriminating based on ethnicity, religion, or otherspecial interests. Many regions have resisted takingon staff transferred from the center, owing largelyto ethnic tensions (central civil servants are prima-rily Javanese) and the association of staff in Jakartawith the authoritarian rule of Soeharto (WorldBank n.d.). Curbing this tendency before it

becomes deeply entrenched will be important, asinterregional mobility of civil servants can helpensure national unity and acceptable levels of serv-ice delivery in more remote regions. Possible mea-sures to address this issue include joint trainingacross regions and a secondment system, in whichfixed-term placements in local areas improve thecareer prospects of central civil servants.

Ironing Out the Wrinkles. Both the Philippinesand Indonesia considered decentralization neces-sary to keep the nation together and regain the trustof disaffected local governments. There was astrong sense that decentralization plans could notawait the relatively long process of institutionaldevelopment and capacity building, and thus thathuman and institutional capabilities should beexpanded during the decentralization process.This approach succeeded in focusing the policyagenda on the actual transfer of functions andmanagement authority to the local level, but left inits wake some unfinished business that hasslowed the momentum behind decentralization.In Indonesia, concern over the loose ends hasprompted the central government to roll backdevolved authority.

Capacity. Decentralization policies in thePhilippines and Indonesia have focused on the low-est tier of local government, but implementation ofdevolved functions is running up against thelimited capacity of staff in these administrations,particularly in planning, budgeting, and financialmanagement. In response, the Indonesian govern-ment has adopted a National Framework forCapacity Building, but will now have to work tofinance and implement this plan. It is also unclearwhether the top-down application of this frame-work will adequately address the capacity-buildingneeds of local governments. In the Philippines,the continuing mismatch between required andavailable skills at the local level points to a need forbetter coordination in training local civil servants.Most activities are now provided in isolation, eitherby individual central agencies or by local govern-ments themselves.

Incentives. The decentralization process has alsoaffected incentive structures at the local level. Boththe Philippines and Indonesia have encountereddifficulties in designing intergovernmental financ-ing mechanisms. The Internal Revenue Allotment


east_129-154.qxd 5/17/05 11:42 AM Page 138

in the Philippines has not followed the assignmentof functions as closely as it should have, floodingsome localities with excess resources while saddlingothers with unfunded mandates. The former situa-tion skews incentives toward overspending andoverstaffing, while the latter discourages long-termplanning and spending on maintenance. InIndonesia, the central government’s historical useof the DAU to cover local salary shortfalls and, now,to fully fund subnational wages creates a perverseincentive for overstaffing regional governments. Inboth countries, performance incentives for civilservants are dampened by compressed salary struc-tures, ineffective monitoring of performance, andweak links between performance and pay. Theseeffects have been compounded by a lack of mobilitywithin the civil service—vertically in the case of thePhilippines, and horizontally in Indonesia.

Autonomy. In its initial thrust, decentralizationin the Philippines and Indonesia significantlyboosted the autonomy of local managers. Indone-sia, however, has recently repealed local staffingauthority. In both countries the central governmenthas retained control in defining pay levels for localcivil servants. Subnational governments are subjectto national wage scales that impose significantfiscal burdens, especially in poorer localities.Centrally imposed salary increases have furtherstrained local budgets. In Indonesia, the DAUcovered these, but in the Philippines, the SalaryStandardization Law is not yet fully implementedbecause many local governments find it simplyunaffordable. Another unfunded mandate thatimpinges on the autonomy of local civil servicemanagers in the Philippines is the 1993 MagnaCarta of Public Health Workers Act. The centralgovernment extended this generous package ofwage and benefit guarantees to health employeesas an inducement to accept devolution. This haswidened an already noticeable gap in the remuner-ation of local versus devolved staff, and hasimposed significant costs on local governmentswhile reducing the ability of local managers to allo-cate resources according to need. In both Indonesiaand the Philippines, limits on local authority haveled managers to bypass established regulations byhiring staff on a temporary basis and topping upsalaries with nontransparent allowances.

Accountability. The tendency to circumvent therules may reflect an adaptive response by local

managers to do what needs to be done, but theresulting lack of clarity and transparency poses aconsiderable risk in reduced accountability and vul-nerability to corruption. In the Philippines, weakcontrols on staff appointments open the door topatronage and nepotism in local government. InIndonesia, though the speed of decentralizationprobably prevented vested interests from influenc-ing decentralization policies, there is evidence thatlocal positions are bought and sold in return for thepromised rewards of graft in the public service(World Bank 2003c). The accountability of localcivil servants is further threatened by a lack ofclarity—in Indonesia because of imprecise roles,and in the Philippines owing to the “double subor-dination” of local staff to the local executive and therelevant central agency. Internal accountabilitymechanisms remain weak in both countries, thoughexternal accountability institutions—such asombudsman institutions, complaint bureaus, andcitizen monitoring—are growing in importance.

Overall, the experience of East Asia’s rapiddecentralizers shows that, although establishing theright rules is important to administrative decen-tralization, good design on paper is not enough.Countries must pay attention to the realities ofimplementation, particularly to creating incentivesfor local governments to work within the rules andto maintaining the momentum to make neededpolicy adjustments.

Decentralization in Transition: China and Vietnam

Administrative decentralization in China andVietnam was born of the transition to a marketeconomy. During the economic opening, centralgovernments relinquished control over some oftheir activities and offered greater administrativeand, to a lesser extent, political independence tosubnational entities in return for greater centralauthority over revenue assignments. This systemhas confused the roles of subnational civil servantsand limited the autonomy of local leaders in man-aging their activities.

On the other hand, in “crossing the river by feel-ing the rocks,”11 decentralization in China andVietnam has encouraged governments to experi-ment with local autonomy and address problemsbefore moving on to the next stage of decentraliza-tion. And local governments in both countries have


east_129-154.qxd 5/17/05 11:42 AM Page 139

formulated innovative responses along the way.The decentralization experiences of these twocountries—different in size yet emerging from sim-ilar bureaucratic traditions—provide insights intothe impacts of administrative decentralization oncivil service management and local service deliveryin transition countries.

The Ebb and Flow of China’s Decentralization. Ina country as large and varied as China, some formof administrative decentralization is a necessity,yet the statutory basis for decentralized humanresource management is fairly limited. The 1993Provisional Regulations on Civil Servants addressthe nationally unified core civil service, whichincludes only white-collar workers such as man-agers and professional staff. Teachers, doctors, sup-port staff, research institute employees, and mem-bers of the military are not considered part of thecore service, but of separate Public Service Units(see below). Political appointees are consideredcivil servants, however, as no distinction is drawnbetween political and bureaucratic personnel.Though the Chinese Communist Party (CCP) isformally separate from the government, it doesinfluence staffing and other aspects of publicmanagement. Central and subnational civil ser-vants alike are subject to national guidelines;although local jurisdictions prepare their own reg-ulations, these usually conform to the nationalrules (Wong 2003).

The People’s Congress usually makes the deci-sion to hire or fire a local civil servant at the corre-sponding level; however, the CCP controls theappointments of senior staff through the nomen-klatura system of bureaucratic patronage commonin many communist countries. Since 1984, high-level officials in each administrative tier have beenappointed by the party committee at the next-highest level.12 Recruitment is supposed to bedetermined by open, competitive examinations,and the selection and promotion of employeesbased on merit and performance. It is important tonote, however, that these criteria are generallydefined to include “political integrity,” or commit-ment to party policies. Local staff are accountableto the central government rather than to localadministrations. Civil servants must follow codesof conduct set out in both the Provisional Regula-tions and party writings. Temporary workers, such

as substitute teachers, are exempt from the nationalguidelines.

The number of posts authorized for local juris-dictions is determined by local branches of theState Commission for Post and Establishment, ajoint government-party organization, at the next-highest level of government. The commission usesa weighted formula to calculate each jurisdiction’soverall staff complement, including a quota for thenumber of support staff.13 Staffing numbers arenot necessarily efficient, however, and inefficienciescan be preserved over time because responsibilitiesfor budget preparation and staffing policy areseparate—a holdover from the days of centralplanning.

Civil servants are paid according to a nationalsalary scale, which is benchmarked to the wages ofstaff with similar responsibilities in state-ownedenterprises. This restriction pegs salaries closer tothose of staff in larger, urban localities. Bonusesand benefits are determined locally and often com-pose a hefty proportion of overall remuneration.Well-off localities use these perks liberally tosupplement the pay of civil service cadres. Poorerareas have difficulty meeting nationally mandatedwage increases, and some have been forced toignore these instructions or supplement personnelbudgets with funds originally allotted to capitalexpenditures. The central pay scale does not, there-fore, ensure nationally consistent pay. The fact thatlocal governments must cover civil service pen-sions, unemployment benefits, and other safety netexpenditures intensifies the fiscal pressures causedby central control over wage levels. When thenational government raises pay, these salary-basedexpenditures rise as well, tightening local budgetsbeyond the control of local managers.

Relatively high salary scales further burden lessprosperous local governments, where the civil serv-ice may be one of few sources of wage employment.Indeed, the tendency for local administrationsto serve as the employer of last resort is exacerbatedby higher-level insistence that they absorb streamsof recent graduates (Wong 2003). These forcestogether inflate staffing levels, especially at thecounty level, which accounts for over half of theexcess staff (World Bank 2002a). Despite severalattempts to downsize the civil service and streamlinethe structure of government, redundant agenciesremain and have tended to staff up again over time.


east_129-154.qxd 5/17/05 11:42 AM Page 140

Wage levels appear to be high enough to allowgovernments to recruit and retain qualifiedemployees. The number of higher-level staff withuniversity degrees has risen from about 20 percentin 1981 to more than 80 percent in 1994. Thoughpay increases and bonuses are designed to reflectperformance evaluations, this does not seem tomotivate staff to perform well. The majority receivepositive appraisals, and managers generally handout bonus payments across the board rather thanto reward exceptional work. Overall pay levels arehighly compressed. The ratio of highest to lowestsalary is 5.6 to 1 for base wages, but flattens tosomething on the order of 3 to 1 when subsidiesand bonuses are included (World Bank 2002a).14

The relatively small increments between pay cate-gories substantially reduce the payoff gained by anemployee who ascends to the next rung of thebureaucratic ladder. There are no explicit restric-tions on the transfer of local civil servants from onelocality to another, or up to the central level, but thestructure of the Chinese labor market—in whichthe hukou system of household registration assignsworkers to a designated area—limits mobility.Some movements occur among higher-rankedstaff, but rarely in the lower echelons. In the late1990s, the central government initiated a programto relocate mid-career officials to local jurisdictionsfor six months to one year as part of a career path tosenior public service.

The core civil service, as defined by the Provi-sional Regulations, does not encompass a substan-tial proportion of government-financed workersemployed by Public Service Units (PSU), such asschools, hospitals, and research institutes. PSUs fig-ure prominently in the public employment land-scape, accounting for 96 percent of the civiliannon-state-enterprise workforce at the central level,and 73 percent at the subnational level. Humanresource management is slightly more flexible inPSUs than in core government entities. Local PSUscan be run at the central or local level or managedjointly. As in the core service, the next-highestbranch of the State Commission for Post and Estab-lishment determines the size of PSU staff. Theserestrictions do not seem to be heeded in practice,however, as local PSU employment averaged 110percent of the authorized staff size in 1999. AndPSU employment is growing, particularly at thesubnational level. In provinces, prefectures, and

counties, PSU staff complements expanded by 21,21, and 75 percent, respectively, between 1991 and1999 (World Bank 2002a).

PSUs are subject to the authority of andaccountable to the central government. The bound-aries between these two forms of public employ-ment often blur. Throughout various attempts todownsize the civil service in the late 1980s and1990s, PSUs often absorbed retrenched employees,limiting cost savings. There is an expectation thatstaff can move freely between PSUs and core gov-ernment positions. In essence, PSUs are a reservoirof extrabudgetary human resources upon whichthe government can draw (World Bank 2002a).15

Moreover, the functions of core agencies and PSUsoften overlap, creating inefficiencies in administer-ing and delivering public services.

Traditional and Innovative Approaches toDecentralization in Vietnam. Along with thetransition to market-oriented economic principles,a major outgrowth of the doi moi reforms begun in1986 has been a shift in the balance of state power,first from the Communist Party of Vietnam (CPV)to the government, and then from central to subna-tional levels. The impetus for administrative decen-tralization is growing with the Master Program onPublic Administration Reform, which stronglyadvocates realigning the management of humanand financial resources to address the fact that“administrative machinery at local levels is notreally responsive to people” (Government ofVietnam, 2001, 2).

Most civil servants in Vietnam are hired andfired by the People’s Committee at the next-highestlevel of government. In the provinces, the Provin-cial People’s Committee makes these decisions,though high-level appointments require the ap-proval of the prime minister. Openings must beadvertised within the locality, usually by way of apublic posting outside the Office of the People’sCommittee. While job classifications have beendefined centrally, they are not accompanied by pre-cise job descriptions, and thus offer local govern-ments significant discretion in determining theirstaffing profile.

All civil service appointments and promotionsare to be based on merit, as determined by exami-nations conducted by provincial-level politicalschools. There are indications that, in practice,


east_129-154.qxd 5/17/05 11:42 AM Page 141

seniority is accorded more weight than perfor-mance in promotion decisions. Selection to higher-level positions also requires ranking membership inthe CPV. No performance management systemlinks high-quality work to high return on a civilservant’s career path. Two 1998 regulations—theOrdinance on Complaints and Denunciations andthe Ordinance on Procedures for Resolving Admin-istrative Disputes—offer mechanisms for disciplin-ing staff, but managers do not exercise either ofthese very often. This is due in the former case to alack of specificity in the provisions, and in the lattercase to the limited effectiveness of AdministrativeCourts. No formal procedures are in place to allowdismissed civil servants to address grievances orappeal personnel actions.

Though officially no geographical constraintshinder the recruitment and transfer of staff, incen-tives for interregional mobility are generally inade-quate, as local residents fill most local positions.Furthermore, remote areas tend to have difficultyattracting high-quality employees. There is alsoroom to boost recruitment of regional minorities,particularly above the commune level. In theory,local civil servants are not barred from moving intothe central civil service, but in practice, only People’sCommittee chairs move up to service in Hanoi.

Instead, central civil servants seem to be spend-ing more time away from the capital. In 2003, thegovernment launched a new program of staff rota-tions aimed at deepening the hands-on experienceof top central bureaucrats. The intent is that timespent at the grassroots level will serve as a criticalstep on the path to high-level government posts(Cohen 2003). Whether this promising initiativewill bear fruit, however, remains to be seen. Thereis concern that the rotation scheme has been usednot as a tool for instilling an improved publicservice ethos but instead to banish politicalopponents.

The central Ministry of Home Affairs managesthe personnel establishment through staff quotas ateach level of government. Until 2004, these staffquotas also determined the budget allocations forpersonnel expenditures at the local level, but theone-to-one relationship between a larger staff sizeand a larger budget allocation created incentives foroverstaffing and penalized leaner, more efficientadministrations. Budgets now reflect a province’spopulation rather than its staffing levels.

Salaries and allowances are specified centrallyand apply uniformly to all levels of government.Wage levels have traditionally been considered lowcompared with alternative sources of employment,though new analysis casts doubt on the validity ofthese comparisons. Bales and Rama argue that,while professional and technical employees maybe underpaid compared with the private sector, thealternative employer for most civil servants—particularly those based in rural areas—is mostlikely the informal sector, where salaries tend to belower (Bales and Rama 2002). In poorer regions, infact, the comparatively high wages fixed by thecenter—along with recent salary increases—aretightening budgets. This pressure will only buildwith the planned pay reform program, whichpromises to raise salaries by 30 percent, on average.Because staff pay represents a considerable share ofspending in service sectors, wage costs can freezeout other important spending. Though the crowd-ing out of operations and maintenance expendi-tures is not a major problem overall in Vietnam, itis an important issue in poorer areas.

Turning Transition into Forward Momentum.Decentralization in China and Vietnam has beenincremental, with the central government reactingto the effects of each policy before taking the nextstep. While this approach allows experimentation toaddress unexpected results before they becomeentrenched, the lack of a predefined plan can createinconsistencies in decentralized governance. Forexample, sufficient fiscal resources have not accom-panied administrative mandates—indeed, in China,the recent trend has been toward recentralization offinances. In Vietnam, the PAR Master Program,which sees the capacity and accountability of civilservants as critical to improving the effectivenessand efficiency of public service, is an important steptoward redressing imbalances in fiscal and humanresources.

Capacity. The PAR Master Program calls fordeveloping a capacity-building plan to provide bet-ter training opportunities at local as well as centrallevels, refresh the curriculum and methods, andalign content more closely with needed on-the-jobskills. Enhancing the reach of educational opportu-nities will be important in addressing the lowcapacity of many local governments in Vietnam,particularly in remote areas. China has also made


east_129-154.qxd 5/17/05 11:42 AM Page 142

training civil servants a priority: local governmentsmust upgrade the skills of civil servants in line withthe needs of the market economy. In both coun-tries, the central government provides resourcesfor locally organized training programs. In China,local governments are taking advantage of theseresources to send their employees to Chinese uni-versities as well as training programs outside thecountry. However, local administrations in bothChina and Vietnam continue to lack expertise inaccounting and financial management. In Vietnam,a recent audit found an absence of record keepingin the communes, which keeps local managers fromeffectively overseeing the civil service and preventsthe center from tracking the use of funds.

Incentives. Improving the ability of local govern-ments to manage financial resources addresses onlypart of the problem when the incentives of civil serv-ice managers are skewed away from prudent super-vision of local budgets. Until recently, Vietnam’ssystem for allocating administrative budgets—calculated per staff member per year—encouragedlocal managers to inflate staffing levels to obtainmore resources from the center. In China, local offi-cials have little motivation to reduce personnel costs,as they are budgeted separately from other expendi-tures. Indeed, managers who save on wages lose partof their budget allocation. In poor regions, addingstaff is often seen as the only way to increase the localbudget allocation. Coupled with the government’sposition as employer of last resort, this system hasencouraged overstaffing in local governments.

The ability of decentralization to improve serv-ice delivery also depends on the incentives of civilservants themselves. In both China and Vietnam,despite official rules defining merit as the basis forpromotion, the career paths of local staff appear tobe more limited than those of their central govern-ment counterparts. Seniority and party loyaltyreceive as much or more consideration than per-formance, and salary compression limits the bene-fits gained from hard work. China’s attempt toenhance incentives for high-quality work throughperformance-based pay has not succeeded becausemanagers have applied bonuses across the board.Staff rotation schemes in both countries focus ondeveloping the careers of central civil servantsthrough time spent in regions, rather than on rotat-ing local staff to other areas to facilitate sharing ofexperiences and information (see box 7.1). The

Vietnamese government has had significant successin improving incentives for prudent managementof administrative expenditures through an innova-tive pilot program for block-grant budgeting in HoChi Minh City (see box 5.3).

Autonomy. Controlling administrative expendi-tures has indeed proved difficult for local managersin these countries because of their restricted auton-omy in setting wage and staffing levels and, inChina, associated safety net expenditures. The highpercentage of local budgets preempted by person-nel costs, in turn, limits local discretion regardingother expenditures. In the meantime, local leadershave focused on a few areas of personnel manage-ment in which they do have autonomy. In China,local governments determine bonus pay, and haveused it to supplement civil service wages. In bothcountries, local governments have the right to sup-plement their budgets by collecting user fees forcertain services. While this has eased the strain onsome local budgets, many of these fees fall dispro-portionately on the poor. Moreover, because theseresources are external to the budget, governmentsoften spend them in nontransparent ways thatdo not reflect budget priorities and undermineaccountability.

Accountability. The accountability of local civilservants in China and Vietnam is blurred by a lackof clarity on the roles and responsibilities of eachlevel of government, though Vietnam’s new Bud-get Law has helped to address this problem. InVietnam, and at the township level in China, thelack of clarity is intensified by dual subordination,in which local staff are beholden to both the direc-tives of the central government and the demands oflocal elected assemblies. Audits of local expendi-tures in both countries have found that accounta-bility mechanisms are weak. In China, the internalaudit function is strong, focusing mainly on ensur-ing compliance and preventing fraud. Externalex post audits are less effective, as local auditbureaus report to the central government and thuslack needed independence. The effectiveness ofaudits in Vietnam is constrained by a lack of capac-ity and low pay. In both countries, efforts are underway to improve accounting and financial manage-ment at central as well as local levels. These mea-sures will help the central government improve andmonitor local service delivery. The presence of theCommunist Party in China and Vietnam holds


east_129-154.qxd 5/17/05 11:42 AM Page 143

important consequences for the accountability ofcivil servants in a decentralized setting. The partyexerts considerable influence over staff appoint-ments and promotions, particularly at senior levels.As such, civil servants most likely feel moreaccountable to the party than to the local popula-tion. While the party can be seen as a source of dis-cipline in an otherwise weak system, its traditionalcontrol over the flow of information can reducetransparency and undercut efforts to hold civil ser-vants accountable (Wong 2003).

Government transparency is beginning toimprove in China and Vietnam, both within the civilservice and with respect to the public. The state-run

media and citizens are paying increasing attentionto corruption issues. In Vietnam, after peasantprotests in Thai Binh and Dong Nai provincesbrought to light the extent of corruption in the localcivil service, the central government introduced aGrassroots Democracy Decree to enhance local con-sultation with residents on decisions that affectthem. While this could in principle provide a strongcheck on abuse of power and strengthen accounta-bility by increasing the demand for good gover-nance, implementation has been weak because oflimited management responsibility at the communelevel, insufficient training of civil servants in sharinginformation with local citizens, and inadequate


BOX 7.1 Personnel Exchange in India and Japan: Whose Capacity IsBeing Strengthened?

In many civil services around the world, systemsenable staff to transfer across levels of govern-ment. The motivations behind these personneltransfers are varied. In some cases, the intent isto groom central civil servants for high-levelpostings by using hands-on work at the locallevel to ground and broaden their policy skills. Inother cases, central government staff are trans-ferred to subnational jurisdictions to fill gaps inlocal capacity to implement national programs.Where civil servants are transferred acrosssubnational governments, the aim may be tokeep them at a safe distance from local vestedinterests.

A comparison of civil servant rotationschemes in India and Japan demonstrates howthe design of these human resource manage-ment tools can shape their impact in a decentral-ized setting. In India, the principal focus is onpreparing members of the elite Indian Adminis-trative Service (IAS) for senior positions in centraland state governments. Those who make the cutin the highly competitive recruitment processand rigorous training program are dispatched tostate governments, where they spend severalyears working in field and secretariat assign-ments in different sectors. After reaching seniorpositions in the field, IAS officers typically rotatebetween central and state governments, asrepeated exposure to field conditions is key tokeeping abreast of changing on-the-groundrealities. The IAS system also fosters active

collaboration between the center and subna-tional governments.

In Japan, on the other hand, the impetus forpersonnel exchange arose from the need toboost technical capacity in newly establishedlocal governments. Many local postings havesince become “hereditary,” in the sense thatstaff members returning to central governmentare typically replaced by other central staff fromthe same ministry. Though it could be arguedthat this practice has prevented qualified localstaff from gaining access to senior positionsin subnational governments, the fact thatautonomous local administrations continue topay for relocated central staff implies that thetransfer of information and knowledge fromthe center still plays a useful role. Moreover, theJapanese system of personnel exchanges ismuch broader than that in India. Staff not onlytransfer from the center to prefectural govern-ments, but from prefectural governments tomunicipalities, from local governments to cen-tral postings, and across prefectures as well.While in both countries placing centrallyrecruited officials in local administrations canstrengthen the capacity of local staff as well astransferred officials, the design of the Japanesesystem is more conducive to developing theskills of local staff.

Sources: Schiavo-Campo et al. 1997; Inoki 2001; Iqbal2001.

east_129-154.qxd 5/17/05 11:42 AM Page 144

mechanisms for tackling corruption in social assis-tance funds.

The Cautious Decentralizers: Cambodia and Thailand

The implementation of decentralization has beenrelatively limited in Thailand and Cambodia. Inresponse to growing demand for improved servicedelivery at the local level, both countries have rec-ognized the importance of devolving some man-agement responsibilities—and the rights thataccompany them—to local governments. Yet con-cerns about the limited capacity of subnationalgovernments have engendered a cautious approachthat has concentrated on developing the legislativeframework and building local capacity to manageand monitor public services. In Thailand, a weak-ening political imperative to decentralize has alsostalled implementation of outlined policies.

In low-capacity countries, a gradual transfer ofresponsibility can be a pragmatic way to ensurecontinuity in service provision and limit the poten-tial for political capture of inexperienced local civilservants. However, gradualism also has its disad-vantages. Piecemeal approaches may lack an overallstrategic focus, leading to a mismatch betweenresponsibilities and resources, and between author-ity and accountability. Moreover, caution can serveas an excuse for inaction on the part of central offi-cials who are reluctant to relinquish their authority.Thailand and Cambodia have faced some under-standable reasons for delay. Yet in taking such along time to establish the building blocks of decen-tralization, these countries have deferred potentialimprovements in service delivery. In the meantime,the success of several small-scale experiments withlocal autonomy indicates that, with careful design,decentralization can work in these countries. Thefollowing discussion of guarded decentralization inThailand and Cambodia explores how countriesmay resolve the tension between exercising cautionand achieving tangible results.

Building the Foundations for DecentralizingPersonnel Management in Thailand. Thailand’s1997 Constitution represents a significant step inbringing government closer to the people in atraditionally centralized country. Building on thissolid foundation, efforts have shifted to creating

the institutions and developing the legal basisfor devolving responsibilities to subnationaladministrations.

So far, few civil servants have been transferredto local administrations. Only about 4,000 staff(less than 1 percent of the 1.2 million nationalcivil servants) in five departments—principally theDepartment of Public Works and the Ministryof Interior’s Accelerated Rural DevelopmentDepartment—have moved to local service. Thetransfer of staff has thus not followed the transferof functions, as stipulated in the decentralizationlegislation.16 The strategy of the Office of the CivilService Commission for transferring remainingstaff prioritizes voluntary transfers, though manda-tory transfers and compensated retrenchment arestill options. The commission has set up a PublicSector Personnel Development and DeploymentCenter as a hub for training and deploying centralstaff to positions in local government. Provincialpersonnel transfer centers are evaluating thestaffing implications of devolved authority for localgovernments.

Still, the low number of relocations thus farpoints to a lack of motivation on the part of bothcentral agencies and public employees. Ministries inBangkok are understandably unenthusiastic aboutletting their staff go. Poorer local governments, espe-cially the small tambons, may be reluctant to acceptmore staff because of the burden they may pose totight budgets.Employees hesitate to move owing to alack of clarity about comparable pay, benefits(including pension fund provisions), recruitmentand dismissal procedures, career mobility, and localquality of life. Some staff worry that relocation tosmaller local communities will leave them—andtheir jobs—at the mercy of local politics (Wegelin2002). Civil servants who transfer to the regions areentitled to the same pay as in the center, but the sys-tem for classifying local positions does not offer thesame range of job levels (and accompanying paygrades).17 The central government has offered afinancial incentive to public servants who volunteerto serve in local governments for a three-year term,but this does not address the longer-term issue ofpermanent personnel transfers.18

The central government exercises generalauthority over personnel decisions at each level ofsubnational government through the Civil ServiceCommission. Local administrations may hire


east_129-154.qxd 5/17/05 11:42 AM Page 145

low-level staff and contract workers but are notauthorized to recruit higher-skilled or careeremployees. Vacancies are not formally advertised.Recruitment is based on competitive examinationsadministered by the commission, although it canwaive the exam requirement in certain circum-stances. All civil servants are subject to a code ofprofessional ethics, which requires those in higher-level positions to publicly declare their assets.Disciplinary procedures are rigorous and providefor appeal.

The center rigidly prescribes the number andgrade level of authorized staff positions at eachlevel of local government. Fifth-class tambons mayemploy no more than 3 staff members, and first-class tambons no more than 21. Limiting staff sostrictly may undercut remote rural areas that needthe most intensive work on service delivery andpublic outreach. Yet keeping staff lean can helplocalities manage the fiscal burden of civil servicesalaries. Local governments must adhere to a cen-tral pay scale and a 40 percent cap on personnelexpenditures as a proportion of local spending.This is a Catch-22 for poorer tambons, where asheer lack of staff constrains the capacity to per-form local functions, while paying more peoplewould leave little money for anything else (Wegelin2002). To address local personnel issues and redressdisincentives to staff transfers, the Local CivilService Administration Working Group has pro-posed creating a Local Government Civil ServiceCommission—as distinct from the provincial andmunicipal commissions that already exist. TheMinistry of Interior is also looking into a personnelmanagement system to track the pay, benefits, andcareer streams of subnational civil servants. Finally,the Thai government’s recent acceleration ofbroader civil service reforms may reinvigorate thedecentralization process.

Top-Down and Bottom-Up Approaches toDecentralization in Cambodia. According toits National Poverty Reduction Strategy for2003–2005, Cambodia is pursuing a two-prongedapproach to the devolution of power, tackling bothpolitical decentralization to the communes andfunctional deconcentration to provincial-level out-posts of the central government. The governmenthas recently initiated a process to develop an inte-grated policy framework.

Cambodia has established the legal frameworkfor devolution, but its impact on staffing has beenlimited. While recognizing that communes shouldhave their own financial resources, the Law onAdministrative Management is much less ambitiousabout decentralizing personnel and humanresource management. The law requires only twotechnical staff positions at the commune level: acommune clerk appointed and paid by the Ministryof Interior to assist the commune council, and anaccountant from the provincial treasury. Theseindividuals wield central influence over the other-wise relatively autonomous commune councils,although the commune chief can request theappointment of a new clerk based on the council’sdecision. If needed, councils may directly employother staff, whom the law deems outside the “Stateframework.”These employees are retained on a tem-porary basis for the duration of the council’s man-date, but the next council can extend their terms.

At the commune level, general administrationexpenditures are restricted to 30 percent or less ofthe allocation from the center. This limit reducesthe temptation for nepotism and overstaffing whilegiving communes considerable discretion in howfunds are spent within the threshold. There is con-cern that the centrally determined allowances paidto commune councilors are high relative to civilservant salaries, especially as the former are onlypart-time functionaries. This can foster resentmentamong local staff and further restrict administra-tive expenditures.

The legal framework for deconcentration is lesswell defined, but the number of staff affected ishigh. As of 2003, some 78 percent of staff werebased in provinces, including health and educationemployees. The relevant central ministry deter-mines staffing needs and positions at the provinciallevel, but provincial administrations have discre-tion in whom they hire (World Bank 2002b).Though recruitment is supposed to be based oncompetitive examinations administered by theState Secretariat for Civil Service, anecdotal evi-dence indicates that provinces do not strictly followthis requirement (World Bank 2003a). Similarly,promotions within each grading category are to bebased on merit, but seniority appears to play astronger role in practice. Performance evaluationsshould be conducted annually, but this is notcommon. Disciplinary procedures exist but are


east_129-154.qxd 5/17/05 11:42 AM Page 146

rarely used. Incentives are therefore not structuredto encourage civil servants to perform well andimprove over time.

Incentives are further dampened by extremelylow and highly compressed pay. Wages are so lowthat many civil servants are forced to seek addi-tional employment in the private and informal sec-tors, and the resulting absenteeism underminespublic sector performance. While ministries areresponsible for monitoring attendance, punish-ment for absenteeism is rare (World Bank 2002b).Low pay also affects how civil servants spend theirtime when they are at work. For example, the pay-ment of per diems for field visits encourages staff tofocus on activities that require travel, regardless ofwhether this is the most efficient use of their timeor is indeed necessary.

The central government determines salaries,allowances, and social benefits, which local govern-ments must pay on a priority basis. This limits theflexibility of provincial budget managers and maycrowd out other important activities. At the provin-cial level, personnel expenditures averaged 48 per-cent of total expenditures in fiscal year 2002,though this figure ranged from 18.6 percent inMondul Kiri to 57.4 percent in Kandal (WorldBank 2002b). This share is much higher than the33 percent that the central government spends onits wage bill (World Bank 2003a).19 Throughout theCambodian civil service, a lack of systematic estab-lishment control has allowed the proliferation ofcontract workers and high levels of overtime pay,and these problems are likely even more noticeableat the local level.

The combination of low pay, low staff motiva-tion, and low mobility has left line ministries des-perate to skew incentives back toward improvedservice delivery, and several ad hoc responses haveevolved. The Ministry of Health has allowed healthcenters to collect user fees and to apply 49 percentof receipts to supplement staff salaries.20 Thisapproach has shown tangible benefits, as the use ofhealth centers has risen by up to 60 percent. Clientsare evidently willing to pay extra for better service.As Turner notes, “The salary supplements meanthat staff are likely to devote more time to their offi-cial duties. When salaries might be only US$15 permonth, an extra US$10 makes a considerable differ-ence” (2002, 357). However, these supplementsmay lead to pay inequalities across regions. The

Ministry of Education, Youth, and Sports has triedto improve the efficiency of education servicesthrough financial incentives, including pay supple-ments for work in remote areas, double shifts ormultigrade teaching, and performance incentives toheads of provincial and district education offices.

Balancing Caution and Action. The measuredapproaches to decentralization in Thailand andCambodia reflect expressed concerns on the part ofboth governments that local administrations andtheir staff do not yet have the capacity to takeresponsibility for service delivery. In Cambodiathese concerns have some basis in reality; in Thai-land, however, they may reflect a weakened politicaldrive toward decentralization. In both cases, policymakers have sought to avoid the potential dangers ofdecentralization by putting capacity building first.

Capacity. Low capacity is certainly an issue inCambodia, where decades of civil war and theatrocities of the Khmer Rouge regime took adevastating toll on the country’s pool of humanresources. Even in Thailand, a middle-incomecountry with a bustling capital city, most local gov-ernments lack the technical skills and institutionalresources to deliver even basic services. Both coun-tries have created decentralized structures withoutdevolving significant authority or responsibility.

Developing the skills of local staff is seen asthe bridge between these initial steps and more sig-nificant decentralization. In Thailand, municipalstaff are regularly rotated to enhance the sharing ofknowledge and experience among subnational civilservants. Various training institutions offer a widearray of courses, but these have not yet been incor-porated into a focused strategy for developing localcapacity in a sustainable manner. Training andtechnical assistance for decentralization are alsogrowing rapidly in Cambodia, building on the suc-cess of the donor-supported Seila Program. A criti-cal challenge for Cambodia is to leverage this posi-tive experience to design coherent and sustainablecapacity-building initiatives that extend to all localstaff. Thailand and Cambodia have also turned todeconcentration as on-the-job training for civilservants—in Thailand, through Local EducationAuthorities and Area Health Boards, and inCambodia, through Health Center ManagementCommittees, which have the authority to supple-ment staff salaries with proceeds from user fees.


east_129-154.qxd 5/17/05 11:42 AM Page 147

Capacity is not just about technical and bureau-cratic skills, however. The institutional and finan-cial capacity of local governments is also crucial. Inboth Cambodia and Thailand, many local govern-ments are simply too small—in geographicalsize, population, staff numbers, and budgets—toexecute their functions. Most of Thailand’s tambonsemploy only a handful of staff and cannot be con-sidered viable units of government, as many areunable to support even a primary school let alonemore technical services such as infrastructure. InCambodia, a maximum of 11 people sit on thecommune councils, supported by two centrallyappointed civil servants.

Incentives. The small size of local governmentshas influenced the incentives of civil servants totake local employment, which has, in turn, affectedservice delivery. Few local civil service positions areavailable, and those that are sit at lower grade levelswith low salaries and limited career paths. Attract-ing highly skilled professionals is difficult, as localgovernments do not have enough work to sustainfull-time technical experts. As a result, central offi-cials in Thailand have strongly resisted transfer tolocal service. The Thai government has attemptedto address this by offering financial incentives tothose who volunteer, but unclear rules andbenefits have prevented this program from havingany significant effect. In Cambodia, where over three-quarters of staff are based in the field, localstaff are reportedly migrating to provincial centersin search of opportunities to supplement very lowpublic sector wages. Local public servants are ineffect being paid full time to work for part of theday. This is a particular problem in health clinics,which despite a 24-hour mandate often remainclosed for most of the day. The government maywant to consider explicitly allowing part-time workin public health facilities so they can ensure contin-uous availability of care.21

Autonomy. In both Thailand and Cambodia, lim-its on personnel expenditures have prevented localgovernments from hiring enough staff to get the jobdone—and even from filling all the positions towhich they are entitled. In Cambodia’s KampongCham province, however, low use of health servicesmeans that even the small numbers of local staffemployed in provincial and district offices are exces-sive compared with the workload (WHO 2003).This hinders the efficiency and effectiveness of

service delivery. Though local managers in these twocountries do enjoy discretion in whom they hire,rural communities do not have access to the highlyskilled staff available in the center. In Cambodia, alack of predictable financing further constrains theautonomy of local leaders. Unsure of when and inwhat quantity resources will arrive, field-basedmanagers are reluctant to make long-term invest-ments, including in human resources. The govern-ment has taken some steps to alleviate this problemby piloting mechanisms for streamlined executionof budgets in priority sectors. One example is thePriority Action Program, which allows local educa-tion authorities to supplement the low pay of theirstaff. Still, in these traditionally hierarchical soci-eties, local managerial autonomy remains limited.

Accountability. The limited autonomy of subna-tional governments in Cambodia and Thailandposes important challenges for the accountability oflocal staff. In both countries, local governmentsat the lowest tier are officially accountable topopularly elected assemblies. Yet, so far, they areresponsible only for a limited set of tasks, and mostdecisions are handed down from higher levels. As aresult, it is difficult for local citizens to hold civil ser-vants responsible for the quality and timeliness oftheir services. Community monitoring is also lim-ited by incomplete dissemination of informationon decentralization plans and transferred responsi-bilities. The flow of information on public prefer-ences is interrupted by the “missing middle” ofdecentralized politics, as provincial authorities inCambodia and Thailand are not elected and thusreport only to the central government. Despite thetime taken by both governments to lay a solid foun-dation for decentralization, many local officials arestill not fully aware of their own responsibilities—and thus the outcomes for which they are account-able. Local standards and benchmarks are not yet inplace to help the Thai and Cambodian central gov-ernments evaluate local performance. In bothcountries, vote buying, corruption, and elite cap-ture may be significant at local levels, posing animportant risk to decentralization plans.

Successfully Managing the CivilService: The Way Forward

East Asia’s experience with administrative decen-tralization highlights both the opportunities and


east_129-154.qxd 5/17/05 11:42 AM Page 148

the challenges associated with devolving authorityfor managing the civil service to lower levels of gov-ernment. As the case studies show, decentralizingthe management of human resources can improvethe responsiveness and resourcefulness of localgovernments. Yet, without careful design, devolu-tion can also bring fiscal imbalances, negativeincentives, and confused accountability at the locallevel. There is no single formula for successful civilservice management in a decentralized context.

Success can be found in various forms, andtactics that work in one context may produce theopposite result in a different environment. Thediverse approaches of the six cases offer lessons forcountries considering decentralizing civil servicemanagement. The relatively smooth start to BigBang decentralization in Indonesia shows theimportance of carefully managing administrativeelements of the transition, as when the central gov-ernment continued to pay the salaries of trans-ferred civil servants on an interim basis. The lim-ited implementation of decentralization policies inThailand shows that having the approach right onpaper does not necessarily lead to getting it rightin practice. Nonetheless, the latter is important,as evidenced by the legal ambiguities and rever-sals arising in the wake of Indonesia’s rapiddecentralization.

The experience of East Asian countries in man-aging the capacity, incentives, autonomy, andaccountability of the subnational civil service canprovide valuable lessons for other countries.The relatively stable transitions to decentralizationin the Philippines and Indonesia show that coun-tries can strengthen local capacity on-the-job.Cambodia’s positive experience with improving theuse and quality of health services validates the roleof positive incentives—such as salary supplementsfrom user fees—in motivating civil servants toperform better. Vietnam’s experiment with block-grant budgeting in Ho Chi Minh City has reducedadministrative costs and overstaffing, showing thatmore local autonomy can improve civil service effi-ciency.22 Finally, Vietnam’s recent passage of a Bud-get Law that clarifies the responsibilities of variouslevels of government, as well as the experience ofthe Philippines in improving performance throughcommunity report cards, shows how efforts bothwithin and outside the government can improvethe accountability of civil servants.

The experiences of East Asian countries alsohighlight several critical dilemmas that govern-ments must address in designing policies and insti-tutions for administrative decentralization:

Centralized Control vs. Decentralized Management

Implicit in the notion of decentralization is thedevolution of some responsibility and authority tolocal levels, which brings decisions on public serv-ice delivery closer to clients while freeing centralgovernment of the day-to-day details of localadministration. On the other hand, retaining somecontrol at the central level can yield substantialbenefits, including national minimum standardsfor service delivery and wider mobility within thecivil service.

It is critical for decentralization policy to achievea rational equilibrium between these opposing butvalid considerations. Decentralization policiesshould not aim to devolve all authority to the locallevel, but rather to redefine the responsibilities of thecenter and local governments. Central authoritiesmust recast themselves as guides, providing localadministrations with a comprehensive, feasiblenational policy framework and enhancing their abil-ity to evaluate local performance. At the same time,policies must align the responsibilities andautonomy of local managers. The roles of local gov-ernments and their civil servants must be clearlyspecified and accompanied by the resources andflexibility to tailor activities to the needs of residents,as well as by mechanisms to ensure accountabilityand prevent proximity from opening the door toelite capture. Clear, locally feasible standards forminimum performance can mitigate the risks ofdevolving autonomy to subnational governments,as can better monitoring of performance throughthe collection, maintenance, and transmissionof information on performance across levels ofgovernment.

Uniformity vs. Unification of the Civil Service

The degree to which countries devolve authorityfor civil service management has implications foruniformity across the civil service. If the centertotally circumscribes the conditions of local publicemployment, administrative decentralization willremain a myth, and its potential benefits unreal-ized. On the other hand, if local managers areentirely free to adjust salary and staffing levels, the


east_129-154.qxd 5/17/05 11:42 AM Page 149

size and composition of local governments—andtheir paychecks—are likely to vary considerablyand may lead to interregional inequalities inemployment conditions.

In its ideal form, a country’s civil service will beuniform but not unified. Public servants will share adistinct national identity without necessarily beingsubject to identical rules on pay and employment.Countries can foster servicewide identity throughjoint training of civil servants working in differentlocalities, or by creating subnational associationssuch as the League of Mayors in the Philippines orprofessional groups that bring together civil ser-vants who confront similar issues across the coun-try. At the same time, some degree of local variationis important. Solutions to inequitable salary differ-entials include allowing local governments to topup centrally defined base wages through user feesand other own-source revenues, as in Cambodia,China, and Vietnam, or having the center set asalary range but allowing local managers toestablish exact salaries. If working conditions varyacross subnational governments, public service

commissions could manage the careers and interre-gional mobility of subnational civil servants. Thesecommissions must have the authority to recruit,transfer, and properly oversee staff, however (seebox 7.2).

Economies of Scale vs. Client Responsiveness

Administrative decentralization also raises ques-tions about the appropriate level of government towhich responsibilities should be devolved. If thegoal is to bring government closer to the people,then countries should hand off public service pro-vision to the lowest tier of government, where civilservants are best placed to understand and respondto the needs of local residents. However, improvingthe responsiveness of local governments throughproximity can disrupt the potential for economiesof scale. Running a government, no matter how bigor small, requires a minimum number of staff—accountants, managers, and secretaries, forexample—and thus a minimum of human andfinancial resources.


BOX 7.2 Lessons from Uganda’s District Service Commissions

Uganda’s decentralization experience is gener-ally considered a success story in terms of itsextent and impact. As in Indonesia and thePhilippines, decentralization in Uganda took onthe characteristics of a Big Bang following an eraof political opening. Along with the launch ofregular local elections and the transfer of broadservice responsibilities and attendant fiscalresources to local governments, the countrytook steps to decentralize the administrativeapparatus. Staff posted to districts formallytransferred to local governments, and separateDistrict Service Commissions (DSCs) were set upto manage human resources in district and localadministrations.

The right of the DSCs to hire, fire, and over-see subnational staff was enshrined in the newConstitution and further detailed in the LocalGovernment Act. DSCs were meant to provideinstitutional protection for the relative auton-omy of district governments in managingsubnational civil servants. However, the systemhas encountered several obstacles. Recruitmentprocedures are slow, performance evalua-tions rarely occur, and the authority of DSCs to

enforce disciplinary decisions is limited. More-over, because the central Public Service Com-mission defines the size of local governmentsand institutions, vets DSC membership, andapproves recruiting standards, the central gov-ernment retains influence over subnational per-sonnel decisions. Finally, the existence of a sep-arate DSC in each district—and the associatedrise in fiscal transfers and local employment—has created political pressure to add newdistricts. This can lead to the proliferation ofsmaller and smaller local administrations andeventually to a loss of scale economies andoperating efficiency.

As East Asian countries grapple with how bestto manage civil servants under decentralization,the Ugandan experience with DSCs—both posi-tive and negative—may provide a useful guide.If designed well, subnational public service com-missions can promote the qualifications and pro-fessionalism of local civil services throughout thecountry.

Sources: Ndegwa and Levy 2003; Evans and Manning2003.

east_129-154.qxd 5/17/05 11:42 AM Page 150

In devolving responsibilities to the local level, cen-tral governments must consider the capacity of smallcivil services to undertake complex technical andmanagerial tasks, and avoid the proliferation of unvi-able local administrations. Small local governmentsgenerally have difficulty paying the salaries ofhighly skilled staff, and those staff are unlikely tohave enough technical work to keep them busy fulltime. Diseconomies of scale in providing publicservices are a particular issue in Cambodia andThailand, where local budgets and staffs are toosmall to viably provide such services. In thesecountries, ongoing boundary reviews may boostlocal technical capacity by consolidating local units,but they have not led to any decisions thus far.Alternatively, local governments could considerjoining forces to provide public services that requiremore sophisticated technical equipment and skills,or that create spillover benefits for other jurisdic-tions. In China and Indonesia, on the other hand,the larger scale of subnational governments limitstheir ability to effectively reach local populations.

In addition to striking a balance in decentraliza-tion design, countries also face important decisionsin sequencing reforms. Administrative decentral-ization entails two main temporal tradeoffs:

Capacity First or Capacity through Decentrali-zation? How best to sequence decentralizationreforms and capacity-building initiatives is a sub-ject of considerable debate. On the one hand, ana-lysts see devolution of responsibility for manage-ment and public services as a necessary step inbuilding individual and institutional capacity. Skillsdevelopment, performance incentives, and innova-tion are encouraged through “learning by doing,”which is generally a more sustainable and individu-alized approach than occasional courses developedwithout close attention to the specific tasks per-formed by civil servants. On the other hand,Prud’homme and others warn of the “dangers ofdecentralization”: the low capacity and inexperi-ence of local civil servants can mean a sharp deteri-oration in the quality and efficiency of public serv-ices (Prud’homme 1995).

The critical challenge is to maintain forwardmomentum while balancing capacity considerations.Asymmetric approaches to decentralization canallow gradual devolution of responsibilities to localgovernments that have demonstrated the capacity

to take on new tasks. Thailand, for example, hasseen some success with ad hoc forms that devolveautonomy to social sector agencies based on “readi-ness” criteria. However, it is important not to allowoverly targeted approaches to delay the realizationof decentralization’s potential benefits. The experi-ence of East Asia’s rapid decentralizers—Indonesiaand the Philippines—shows that countries canbuild capacity through decentralization. In this sce-nario, careful planning for coordinated trainingwithin a national framework that is sensitive tolocal considerations is essential.

Civil Service Reform before or after Decentrali-zation? Though a full discussion is outside thescope of this chapter, the nexus between decentral-ization and civil service reform is worth mention-ing. In principle, it makes more sense to decentral-ize the structure and management of the civilservice after ensuring that it is efficient and effectiveat the national level. Yet given the political drive fordecentralization and the difficulty of civil servicereform, many countries cannot wait for these tooccur before decentralizing. On the other hand, thedecision to delay civil service reform until afterdecentralization can allow the central governmentto transfer the political burden of difficult policymeasures. Countries should explicitly consider thelinks between these two key reforms to avoid repli-cating national flaws—such as duplication of effortor overstaffing—at the local level.

Endnotes

1. This section draws on Evans and Manning 2003.2. Vaclav Havel, former president of the Czech Republic, used

this phrase to describe the Czech reform experience.3. The Local Government Code of the Philippines, Section 2,

Sections 76 to 97, address human resource managementand development at the local level.

4. The exception to this general rule is the local treasurer, whois appointed by the Department of Finance but paid by thelocal government. Manasan n.d., 12.

5. Philippine Constitution, Article IX, Section 2 (2).6. The 45 percent cap applies in first-class through third-class

local governments, and the 55 percent cap in fourth-classthrough sixth-class local governments.

7. Owing to the penchant among local governments to usealternative sources to supplement wages, these official figureslikely underestimate the degree to which local governmentshave exceeded restrictions on personnel expenditures.

8. Pay scales were set at a fixed percentage of national levels,ranging from 75 percent in sixth-class local governments to95 percent in second-class local governments.


east_129-154.qxd 5/17/05 11:42 AM Page 151

9. Under this revised law, provincial governors will haveauthority to hire, fire, and relocate provincial staff only atechelon 2 and below. Similarly, heads of districts and citieswill be able to hire, fire, and relocate staff only at echelon 2and below, and only after consulting with the provincialgovernor.

10. Some local governments, though, have begun to informallysupplement salaries. Also, several regional allowances areoffered that are imperfectly recorded.

11. At the 11th Party Congress in 1978, Deng Xiaoping usedthis analogy to urge Chinese policy makers to take a grad-ual approach to economic reform.

12. Before 1984, these appointments were made by partycommittees two levels up in the territorial hierarchy. Burnsn.d., 8.

13. At all levels, the criteria used are population, land area, andvalue of industrial and agricultural production; at the citylevel, additional factors include the number of componentadministrative units, local budget income, and the amountof developed land.

14. Including rank-based benefits, such as housing and carallowances, does widen the gap.

15. In Hebei province, for example, the director of budgetpreparation had kept his post as director of the BudgetDepartment in the Hebei finance bureau.

16. Webster (2002) argues that functional devolution does notrequire the transfer of all corresponding civil servants. Heargues that if decentralization is meant to improve the effi-ciency of public services by transferring authority to unitsclosest to the beneficiaries, then fewer people should beable to perform the same task just as well as or better thanat the center.

17. Teacher pay, for example, could be much lower afteremployees transfer to the local level. See Wegelin 2002, 23.

18. The amount of this incentive is B 6,000 per month, orroughly US$150.

19. The national wage bill includes defense and security expen-ditures.

20. The other 51 percent is to be spent on operationalexpenses. See Turner 2002, 357.

21. Conversation with Pamela Messervy, World Health Organi-zation, Phnom Penh, Cambodia, October 2003.

22. See box 5.3. Most surplus staff were transferred to otherlocalities rather than retired or retrenched.

Bibliography

Asia Foundation. 2002. First Indonesia Rapid DecentralizationAppraisal (IRDA) Synopsis of Findings. College Park, MD:University of Maryland, IRIS Center.

Asian Development Bank. 1999.“Governance in Thailand: Chal-lenges, Issues, and Prospects.” Manila: Asian DevelopmentBank.

Azfar, Omar, Tugrul Gurgur, Satu Kähkönen, Anthony Lanyi,and Patrick Meagher. 2000. “Decentralization and Gover-nance: An Empirical Investigation of Public Service Deliveryin the Philippines.” College Park, MD: University of Mary-land, IRIS Center; and Washington, DC: World Bank.

Bales, Sarah, and Martin Rama. 2002. “Are Public Sector Work-ers Underpaid? Appropriate Comparators in a DevelopingCountry.” World Bank. http://econ.worldbank.org/files/3358_wps2747.pdf.

Burns, John. “Country Reform Summaries: China.” www1.worldbank.org/publicsector/civilservice/rsChina.pdf.

Cohen, Margot. 2003. “Cadres Get a Field Lesson.” Far EasternEconomic Review 166 (45): 24.

Evans, Anne, with Nick Manning. 2003. “Decentralization: AReview of Staffing Practices in Eight Countries.” Internaldraft document. World Bank, Washington, DC.

Fritzen, Scott. 2002. “The ‘Foundation of Public Administra-tion’? Decentralization and Its Discontents in TransitionalVietnam.” Paper presented at the Asia Conference on Gover-nance in Asia: Culture, Ethics, Institutional Reform, andPolicy Change, City University of Hong Kong.

Government of Vietnam. 2001. “Master Programme on PublicAdministration Reform for the Period 2001–2010.”Attachment to the Prime Minister’s Approval DecisionNo. 136/2001/QD-TTg, September 17.

Government of Vietnam, General Statistical Office. 2003. Statis-tical Yearbook 2002. Hanoi: Government of Vietnam.

Government of Vietnam, Ministry of Finance. 2000. “Vietnam:Improving the Budget-Norm System for Recurrent Expendi-tures.” VIE/96/208, Public Expenditure Review Phase II.www.undp.org.vn/projects/vie96028/norms.pdf.

Inoki, Takenori. 2001. “Personnel Exchange among Central andLocal Governments in Japan.” Working Paper. Washington,DC: World Bank Institute.

Iqbal, Farrukh. 2001. “Evolution and Salient Characteristics ofthe Japanese Local Government System.” Working Paper.Washington, DC: World Bank Institute.

Kingdom of Thailand. 1997. Constitution of the Kingdom ofThailand. www.oefre.unibe.ch/law/icl/th00t___.html.

Litvack, Jennie, Junaid Ahmad, and Richard Bird. 1998.“Rethinking Decentralization in Developing Countries.”Sector Studies Series. Washington, DC: World Bank.

Litvack, Jennie, and Jessica Seddon. 1999. “DecentralizationBriefing Notes.” Working Paper. Washington, DC: WorldBank Institute.

Manasan, Rosario G. “Fiscal Decentralization: The Case of thePhilippines.” World Bank. http://www.worldbank.org/wbi/publicfinance/documents/ASEM/manasan.pdf.

Ndegwa, Stephen, and Brian Levy. 2003. “The Politics of Decen-tralization in Africa: A Comparative Analysis.” Paper pre-sented at the Regional Workshop on Governance and PublicManagement, Johannesburg.

OECD (Organisation for Economic Co-operation and Develop-ment). 1997. “Managing across Levels of Government.”Paris: OECD.

People’s Republic of China. 1982. Constitution of the People’sRepublic of China. http://english.peopledaily.com.cn/constitution/constitution.html.

Prud’homme, Rémy. 1995. “The Dangers of Decentralization.”World Bank Research Observer 10 (2): 201–20.

Republic of the Philippines. 1987. 1987 Constitution of thePhilippines. www.chanrobles.com/philsupremelaw1.htm.

———. 1991. The Local Government Code of the Philippines(Republic Act No. 7160). www.chanrobles.com/localgov.htm.

Royal Government of Cambodia. 2002. National Poverty Reduc-tion Strategy, 2003–2005. Phnom Penh: Royal Governmentof Cambodia.

Schiavo-Campo, Salvatore, Giulio de Tommaso, and AmitabhaMukherjee. 1997. “Government Employment and Pay: AGlobal and Regional Perspective.” Policy Research WorkingPaper No. 1771. Washington, DC: World Bank.

Steedman, David W., and Lloyd R. Kenward. 2005. “Civil ServiceReforms at the Regional Level: Opportunities and Con-straints.” Internal draft document. World Bank, Washington,DC.


east_129-154.qxd 5/17/05 11:42 AM Page 152

Turner, Mark. 2002. “Whatever Happened to Deconcentration?Recent Initiatives in Cambodia.” Public Administration andDevelopment 22: 353–64.

Webster, Douglas. 2002. “Implementing Decentralization inThailand: The Road Forward.” Internal draft document.World Bank. http://www.worldbank.org/wbi/publicfinance/documents/East%20Asian%20Decentralization%20Dialogues%20II/Douglas.pdf.

Wegelin, Emiel. 2002. “Thailand: Decentralization CapacityAssessment, Findings, and Recommendations.” Internaldraft document. World Bank. http://www1.worldbank.org/wbiep/decentralization/eaplib/wegelin.pdf.

Weist, Dana. 2003. “Thailand’s Decentralization: Progress andProspects.” Paper prepared for the Asian Development Con-ference on Decentralization in Asia, Kita-kyushu, Japan.

Wong, Christine. 2003. “Country Review: The People’sRepublic of China.” Internal draft document. World Bank,Washington, DC.

World Bank. 2000. “Vietnam: Managing Public ResourcesBetter.” Public Expenditure Review. Washington, DC: WorldBank.

———. 2001. “Indonesia: Priorities for Civil Service Reform.”Internal draft document. World Bank, Washington, DC.

———. 2002a. “China: National Development and SubnationalFinance—A Review of Provincial Expenditures.” Washing-ton, DC: World Bank.

———. 2002b. “Decentralization and Deconcentration inCambodia.” Internal draft document for the CambodiaPublic Expenditure Review. World Bank, Washington, DC.

———. 2002c. “Vietnam: Delivering on Its Promise.” VietnamDevelopment Report. Washington, DC: World Bank.

———. 2003a.“Cambodia: Enhancing Service Delivery throughImproved Resource Allocation and Institutional Reform.”Integrated Fiduciary Assessment and Public ExpenditureReview. Washington, DC: World Bank.

———. 2003b. “China Country Economic Memorandum: Pro-moting Growth with Equity.” Washington, DC: World Bank.

———. 2003c. “Combating Corruption in Indonesia: Enhanc-ing Accountability for Development.” Washington, DC:World Bank.

———. 2003d. “Decentralizing Indonesia. A Regional PublicExpenditure Review Overview Report.” Washington, DC:World Bank.

———. 2003e. “Human Resource Management.” Internal draftdocument for Philippines Decentralization and ServiceDelivery Study. World Bank, Washington, DC.

———. 2003f. “Philippines Decentralization and ServiceDelivery Study.” Internal draft document. World Bank,Washington, DC.

———. 2003g. “Thailand Country Development Partnership:Governance and Public Sector Reform,” Phase I MonitoringWorkshop. Washington, DC: World Bank.

———. 2003h. World Development Report 2004: Making ServicesWork for Poor People. Washington, DC: World Bank; andNew York: Oxford University Press.

———. n.d. “Subnational Government.” www1.worldbank.org/publicsector/civilservice/subnational.htm.

WHO (World Health Organization). 2003. “Functional Analysisfor Human Resources: Kampong Cham Province.” Geneva:World Health Organization.


east_129-154.qxd 5/17/05 11:42 AM Page 153

east_129-154.qxd 5/17/05 11:42 AM Page 154

Documents

managing human resources in a decentr alized context