Public Finance, Security, and Development: A Framework and an Application to Afghanistan

Security is increasingly viewed as a key condition for economic growth and development. The authors argue that the work and impact of all development partners would be enhanced if the multiple linkages between public finance, security, and development were explicitly taken into account. At the extreme, in some cases better public finance management could have more impact on security than would more troops. The paper first outlines three core linkages between security and development - through the investment climate, human and social capital, and institutions. The authors then propose three complementary tools to analyze the security sector from the point of view of public finance management, service delivery, and governance. This conceptual framework is applied to the case of Afghanistan. The paper closes by drawing some conclusions about possible entry points for dialogue in this difficult area.


Introduction
It has been estimated that more than 800,000 people around the world lose their lives to lethal violence every year. Almost one-third of all developing countries are considered to be at risk of violent conflict or in a state of chronic instability. Private firms report that they pay up to 15% of their sales revenue to protect themselves from insecurity. 2 report on crime and development in Africa (UNODC, 2005). The UK Government's Department for International Development (DFID, 1999 and has prepared a document entitled "Fighting Poverty to Build a Safer World: a Strategy for Security and Development" (2005). The Development Assistance Committee of the OECD also has issued guidelines on the security-development nexus. USAID (2002), Canada, and others have released policy papers dealing with this same issue as well. Duncan and Pollard (2002) also view social order as an important precondition for growth and poverty reduction, whereas Dixit (2004) provides some counter-arguments, putting forward situations where economic activity can prosper despite "lawlessness". Discussions of security and public finance management are much less common. Notable exceptions include Ball (1988) and Ball and Holmes (2002), the latter with a specific focus on defense. Roberts (2007) is concerned with transparency in the security sector. At the country level, these approaches have only very rarely been used in analytical work. The case of Afghanistan is dealt with in detail in World Bank (2005b, notably in Volume V). In Section 2 of this paper, we synthesize key linkages between security and development. Based on this framework, we then apply three standard public finance tools to the security sector. The central message is that, by and large, these frameworks are very useful for analysis and policy recommendations for the security sector just as they are for other sectors, effectively undermining the common claim that the security sector is special and cannot be assessed like any other sector. Section 3 builds on the three core principles of public finance-fiscal discipline, allocative efficiency, and operational efficiency-to draw lessons for the security sector. Section 4 explores the concept of security as a public service to be delivered to the population and assesses the implications of a service delivery model for the security sector. Section 5 reviews the implications of various public finance management and governance issues for the security sector. Section 6 applies these frameworks to the case of Afghanistan, and Section 7 concludes the paper by discussing possible entry points for dialogue on the security sector from a public finance and development perspective.

Insecurity as a Constraint to Growth and Development
In this section, we propose a conceptual framework to explore the relationships between security and development. We first define the notion of security. The importance of security is typically most evident when it is absent, so security has often been defined in contrast to its opposite, insecurity. From a more positive perspective, the OECD's Development Advisory Committee (DAC) defines security as a development objective in the following way: "Security is increasingly viewed as an all-encompassing condition in which people and communities live in freedom, peace and safety, participate fully in the governance of their countries, enjoy the protection of fundamental rights, have access to resources and the basic necessities of life, and inhabit an environment which is not detrimental to their health and well-being. The security of people and the security of states are mutually reinforcing. A wide range of state institutions and other entities may be responsible for ensuring some aspect of security." (OECD, 2005b, pp. 20-21). This definition is consistent with the broad concept of "human security" used by the UN.
For analytical purposes, we focus on a well-defined set of issues that can be defined by the functions or by the risks covered by our analysis. 3 We approach security in terms of risks, including (Table 1): physical (risk of death or injuries); social (e.g. crime); or political (e.g. coup d 'etat). The implications of these risks also differ depending on whether they occur mainly at the individual/household (micro), local/regional (meso), or national (macro) level.  Holzmann and Jørgensen (2000), which also includes other risks (e.g. natural and economic risks).
The question is then how to manage these security-related risks. In this context of risk management, the framework developed in the seminal work of Holzmann and Jørgensen (2000) can be used. A fruitful distinction in this framework is between risk reduction (reducing the probability of a downside risk), mitigation (reducing the impact of a future downside risk), and coping (relieving the impact of a risk after it has materialized). This highlights that risks have a direct impact (for example, destruction of irrigation systems during a war has a direct cost in terms of reduced economic activity and loss of economic assets), an indirect impact due to coping strategies (e.g. selling livestock to cope with the economic loss, or leaving the country), and an indirect impact due to ex-ante risk reduction or mitigation measures (e.g. reducing irrigation investment due to fear of losing such assets). Importantly, these three impacts have both static (e.g. loss of life today) and dynamic (e.g. reduced human capital in the future) effects. We posit that these various impacts can be summarized around three main channels, which encompass key drivers of growth and development: investment climate, human and social capital, and public institutions (see Aghion and Durlauf, 2005). From the Holzmann and Jørgensen framework, we can see that security-related risks have an impact on growth and development not only directly, but also through the way individuals, communities, and firms respond ex post when risks materialize (coping strategy) and ex ante to reduce the likelihood or magnitude of these shocks (Table 2). In this section, we review the key transmission links at play for each channel and suggest ways to assess their importance in a given country. 3 Although we are trying to provide a holistic view of the security -development nexus, some linkages are excluded. First, we do not consider the global level, which has received much attention after the terrorist attacks of September 11, 2001. Second, and related, we are focusing on policy implications at the national level, not on international institutions or cooperation. Also, our focus is on developing countries. Finally, we do not consider some of the impacts of poor security on development more broadly defined, for instance on environment, migration, and youth.

Security and the Investment Climate
Lack of security weakens the investment climate (World Bank, 2004). This is obvious in the case of conflict, which not only reduces the incentives for investment but also weakens physical capital (through direct damage, or by diverting funds from operation and maintenance), financial capital (capital flight), human capital (again through direct impactsmorbidity, health status-or by indirect impacts such as reduced supply of education and health services, outward migration of refugees), and social capital (disruption of communities, involvement of men in conflict, worsening mistrust between fighting factions). Conflict also disrupts trade and markets, leads to theft of business property by armed groups, and imposes "protection costs" on surviving firms. Most of these costs tend to continue for an extended period after the end of the conflict (infrastructure takes time to be rebuilt; improving education and health outcomes, as well as social capital, is a slow process; capital flight tends to continue or even increase after conflict). Criminal activities can weaken the investment climate. They are often directed against businesses and hence directly inflict losses on them, but more damaging is the harm crime and perceptions of it do to business investments. Foreign investors may be discouraged, domestic investors may not start activities, and financial capital may flee from high-crime areas and countries. In addition, insecurity can generate macroeconomic instability, also weakening the investment climate and further fueling risks of conflict (World Bank, 2003a). The indirect costs, related to mitigation strategies, also are large. Criminality or risk of conflict increases uncertainty, to which entrepreneurs respond with more risk-adverse strategies. 4 They also can drive business activity into the informal sector, and keep most businesses small (so as to avoid visibility vis-à-vis criminals or corrupt officials). As a mitigation strategy, firms may have to pay for private modes of providing security. For instance, the World Bank-sponsored Investment Climate Surveys quantify the resources paid by firms for security purposes (Figure 1). Obviously, this is a fraction of the true cost of lack of security given the other indirect impacts listed above. As is discussed below, insecurity also weakens public institutions, opening up vulnerabilities to corruption and further weakening the investment climate. These effects on the investment climate have adverse consequences for investment, technology, productivity growth, and more generally the competitiveness of the economy, reducing prospects for sustained rapid economic growth which is the only way countries can develop and reduce poverty. But in addition, this in turn has a detrimental impact on the capacity of the government to provide security-because the government cannot tax the informal sector or because the willingness to pay taxes has been eroded by corruption-and hence on security itself. 5 Finally, these various costs, including lower economic growth, tend to fall disproportionately on the poor (Box 1).

Box 1: Cost of Conflict and Insecurity for the Poor, Women, and Children
The adverse effects of conflict and insecurity, and often of crime as well, tend to be disproportionately felt by the poor, minorities, and other vulnerable groups. This is partly because as in the case of many other public goods, the better-off can use their own resources to insulate themselves at least to some extent from insecurity (e.g. risk reduction / mitigation strategies such as private security, physical protection through fences etc., mobility), whereas minorities and vulnerable groups are at higher risk of being targeted in situations of conflict or breakdown of law and order when the state is no longer able to protect its citizens. To be sure, where insecurity is generalized and long-lasting, the effects reach all levels of society, and many assets of the rich may get lost or devalued. Nevertheless, the poor are invariably more vulnerable to conflict and insecurity as they have fewer assets and weaker coping mechanisms to start with. Even if they lose many of their assets, better-off people have a greater chance to escape conflict and tend to be in a better position in terms of livelihoods where they flee to. Conflict and insecurity can have especially damaging effects from a gender perspective, notably in the form of escalated violence against women. Rape and sexual assault against women of opposing groups in community-based conflicts are frequently used as an ultimate means of dishonoring entire communities and reducing people's capacities to resist military advances, causing serious social and psychological disruptions. Any children resulting from such encounters will often be rejected by society and lead a pariah's existence. In addition to violence directly against women, the environment of insecurity associated with conflict or a post-conflict situation can have a disproportionate adverse effect on women, as women's greater vulnerability in conflict may lead to imposition of stricter social norms resulting in limited or no access to education and basic health care due to their reduced mobility. Prolonged periods of conflict, high levels of social violence, and lack of security can also be reflected in increased levels of domestic violence, in which women and children are the primary victims. Conflict and insecurity have profoundly damaging effects on children. The loss of one or both parents, homelessness, the need to shoulder economic responsibilities in childhood, exposure to violence and abuse, disruption of elementary education, and lack of access to health services and immunization are notable examples of conflict-induced consequences which have profound and life-long impacts on children.
A way to measure the degree to which the security constraint is binding consists of looking at its price (Hausmann, Rodrik, and Velazquez, 2005). From this perspective, high payments for security protection constitute additional costs of doing business, constraining growth because as a result entrepreneurs cannot fully appropriate the profits from their investments. Symptoms of security-related risks constraining growth include large unofficial payments, a repressive taxation system, and macroeconomic instability. Hausmann et al argue that policy makers should focus on the binding constraint, as removing other constraints might not only fail to generate the desired improvement in the investment climate, but could even have a negative impact. Such analysis of the investment climate could therefore guide the prioritization of efforts across security areas (see below); it would also bring out underlying governance weaknesses which often are behind the manifestations of insecurity.

Security, Human and Social Capital, and Poverty
Human and social capital costs, in addition to their impact on the investment climate, have a direct negative impact on well-being (see for example World Bank, 2002). These factors also generate dynamic effects-either to cope with the risks or to mitigate them-that further increase their costs over time. Across the board, these various costs tend to impact disproportionately on vulnerable parts of the population, such as the poorest, ethnic minorities, and women (Box 1), exacerbating poverty. Immediate impacts include human costs such as deaths, wounds, displacement, and flight, as well as loss of private property and psychological damage. Moreover warfare, especially internal conflicts like civil wars and insurgencies, tends to generate high casualties and other costs among the civilian population. At the individual level, insecurity increases uncertainty and vulnerability, making it more difficult for individuals to plan and invest for the future and to smooth consumption. It also limits the availability of informal risk management mechanisms, when for instance access to family or kin is disrupted by conflict. By limiting movements of people, crime may also impede access to employment and educational opportunities (particularly in the case of females). 6 Finally, ongoing conflict or widespread crime weakens the constraints on criminal and opportunistic behavior. These costs are compounded by a set of dynamic effects stemming from coping with risks once they have materialized. Disruption of public services like education and health due to conflict or insecurity results in lower levels of human capital (due to disruption of education) and higher rates of mortality and morbidity (aggravated by the direct adverse impact of conflict-related phenomena like displacement on health). Human capital can be weakened by the direct effects of crime on victims, and by the associated atmosphere of fear that may encourage many to leave (particularly better-off and skilled people). A need to smooth consumption in the midst of conflict or insecurity can lead to use of costly informal coping mechanisms-such as selling assets (e.g. livestock), reducing risk-taking behavior (e.g. by shifting production to subsistence crops to avoid uncertainty related to market access), shortening time horizons, or turning to illegal activities (e.g. growing illicit crops or engaging in smuggling). Another set of dynamic effects results from mitigation strategies that may be implemented before risks materialize. Emigration and risk-adverse behaviors on the part of those staying in the country lead to less investment in health, education, and innovation, all weakening the country's human and social capital. The poor and disadvantaged have fewer options to mitigate or cope with these risks: hence they tend to be more often and more greatly affected by these risks (Box 1). Informal provision of security by or on behalf of local communities, a practice which often arises when governments fail to provide security, carries a risk that such mechanisms get captured by political or criminal interests, and that the "summary justice" often provided itself can heighten personal insecurity. As in the case of investment climate issues, there are tools to gain a better understanding of some of these effects. A number of poverty assessments, and the underlying household surveys, seek to measure the vulnerability of people, poor and disadvantaged people in particular, to a variety of risks. Although security-related risks are often less straightforward to define and discuss than, say, a climatic risk such as drought, this issue can be analyzed. For instance, the National Rural Vulnerability Assessment in Afghanistan asked households if they had suffered from violence over the previous twelve months (see also Dercon, 2005).

Security and Public Institutions
Finally, lack of security has a very harmful impact on public institutions. First, it adversely affects public finances by directly reducing the capacity to raise revenues if the government does not control the country, and raising expenditures on the security sector when insecurity is rampant. Conflict also destroys state assets (direct destruction; erosion due to lack of maintenance; destructive conflict between factions contesting State authority). Second, by demonstrating that the government has lost the monopoly over organized violence, a key function of the Weberian State, lack of security weakens the legitimacy of the State and hence its capacity to deliver (legitimacy tends to make it easier and cheaper for the State to collect revenues and deliver services). Third, crime can become closely linked to corruption (itself a form of criminal activity), with very destructive consequences for development and public finance management. Corruption and other forms of crime can erode the revenue base of the state and thereby its effective functioning. As a consequence of these effects, trust between people and the state is destroyed or is prevented from being rebuilt. Security-related risks also generate suboptimal ex ante risk mitigation and prevention strategies by public institutions. A classic example is the secrecy around security institutions, often viewed as a necessary working requirement for these institutions, whereas experience in both developed and developing countries indicates otherwise (Roberts, 2007). Policies might also respond in a suboptimal way to a large informal sector, for instance with excessive reliance on trade taxes which are easier to collect than consumption taxes (Emran and Stiglitz, 2005). Anti-corruption strategies also can become counterproductive if excessive controls and untargeted prosecutions lead to paralysis of the State, or if anticorruption is approached by systematically disengaging the State (Khan and Gray, 2006). Similarly, coping strategies are costly. The imperative of revenue collection can lead to excessive enforcement that further deters private investors. Widespread corruption encourages private deals with corrupt officials. Government policies become oriented excessively toward the short-term. Unfortunately, the adverse implications of insecurity for development and for public institutions continue, and in some respects may increase, after the end of outright conflict. So-called "legacy" costs of conflict, based on cross-country analysis (see World Bank, 2003a, pp. 19-31), include in particular (i) higher residual military expenditure; 7 (ii) government policies, which on average are negatively affected after conflict; (iii) deterioration of political institutions which takes time to reverse; and (iv) destruction of the revenue apparatus and loss of revenue culture, which tends to persist after conflict. These legacy costs are often associated with constituencies created during the conflict. For instance, higher military expenditure is usually controlled by interest groups of government officials and military commanders that have little incentive to reduce the size of the army. Corruption can also result in the State degenerating into a "mafia state" where the interests of corrupt officials are closely aligned with those of a corrupt private sector. Weakened (possibly autocratic) political institutions have very limited incentives to reform themselves toward a more participatory and citizen-based approach. For instance, Kaufman (2005) notes that the proportion of firms reporting high costs of organized crime in countries with low institutional capacity of the parliament or police is two to four times higher than is reported in countries with high capacity.
A diagnostic of the impact of security risks on public institutions requires reviewing issues of public finance (revenue collection and allocation of resources to the security sector); political institutions (and accountability to the public at large); and corruption (which requires also looking at the quality of public administration). Elements of such a diagnostic are presented in Sections 3 to 5. Finally, it must be recognized that in addition to the causal link from insecurity to weak and problematic public institutions as described above, the reverse causation also occurs: worsening and weakening public institutions can promote, or at least leave greater scope for, greater insecurity. Moreover, stronger but repressive and predatory states generate insecurity for their populations and can be the main source of human insecurity in some countries. While such states may be able to prevent outright conflict in the short run through repression, experience in some countries suggests that this may build up tensions over time, sometimes resulting in much greater conflict and insecurity when the repressive regime in the end fails. While it is beyond the scope of this paper to explore these issues in detail, the complex and two-way causation between security and public institutions must be kept in mind.

Summing Up
There are two lessons from this review of the linkages between security and development.
The first lesson is that security very much matters. Lack of security directly weakens the investment climate, erodes human and social capital, and weakens public institutions. Governments must therefore devise strategies to improve security. However, there is also a second lesson: security cannot be seen in isolation, for two reasons. First, abolishing security-related risks is impossible (or too expensive for available resources). Hence a security strategy must be cognizant of the private responses to security risks. Lack of security leads individuals, communities, and firms to take actions to reduce the likelihood or potential impact of risks (before they actually occur) and to cope with risks once they materialize. There is a role for government when these actions can themselves have a negative development impact (typically because they use suboptimal informal devices). Second, there are mutually reinforcing effects between security on the one hand and the investment climate, human and social capital, and public institutions on the other. Hence there can be virtuous or vicious circles between security and these three development objectives ( Figure 2). In essence, security risks can be reduced and mitigated through a better investment climate-economic activity generates revenues for the government, which creates fiscal space to provide security services, and reduces the profitability of illegal activities relative to legal activities; stronger human and social capital-trust in the polity facilitates conflict resolution; and effective institutions-they provide the required accountability framework for proper decision-making and policy implementation in the security sector, and they also provide more effective conflict resolution mechanisms. As noted in World Bank (2003a), the probability of civil war tends to decrease with faster economic growth and stronger institutions, which in turn require a base of good governance.

Public Finance in the Security Sector
In this section we apply a public finance framework to the security sector, which considers three key levels of fiscal outcomes: aggregate fiscal discipline, allocative efficiency of fiscal resources, and operational or technical efficiency (World Bank, 1998). Using this framework, we show that, by and large, principles of good public finance management can and should be applied to the security sector, at all three of these levels.

Aggregate Fiscal Discipline
The objective of fiscal sustainability is founded on two rationales. First, to make the budget an effective policy instrument, budgets have to be realistic. Otherwise they become irrelevant since they are never implemented. Second, there is solid evidence that poor fiscal policy-notably in the form of large budget deficits for unproductive purposes or financed by loose monetary policy-has a negative impact on inflation and growth, and hence on poverty. In fact, since there is also evidence that macroeconomic instability can be a source of conflict (World Bank, 2003a), fiscal discipline can be viewed as a necessary condition for a country to achieve its security objectives. Finally, a focus on fiscal discipline is critical since an obvious risk mitigation strategy would be to build very large security forces, whichbeyond the potential governance and security (e.g. coups d'etat) related issues-would jeopardize fiscal stability and hence macroeconomic stability, thereby defeating the risk mitigation strategy. Applying this principle generates three lessons. First and foremost, and applicable to prioritization of spending within the security sector (see below) as well as aggregate security sector expenditures, the security budget and the budgets of its main components (defense, interior, justice, etc.) should be prepared against clear strategies-at the overall, sector, and subsector levels-and corresponding policies. The security strategy in turn needs to be an integral part of the broader national development strategy. These principles are applicable to all parts of the public sector, and security should be no exception. This issue is reviewed in more detail in the next section. Second, as for the budget as a whole and as with any other sector, security sector policies and expenditures must be affordable, both in the short run (annual budget) and in terms of their implications for spending in future years (e.g. decisions on staffing levels, associated non-salary costs, implied pension liabilities, procurement of equipment, and other sustaining costs). This implies that the security sector needs to be fully incorporated in the annual budget formulation process, subject to aggregate fiscal constraints and sector ceilings like any other sector. It also means that the security sector must be fully incorporated in mediumterm fiscal projections and planning. Finally, all government transactions in the security sector should be on-budget, including revenues earned directly by security sector entities. Third and equally important is the need for transparency and information, without which decision makers cannot make informed choices with respect to the budget. Legitimate issues of confidentiality may constrain public dissemination of some information, but this should not prevent sufficient relevant information from being made available to decision makers, if necessary through devices like closed budget hearings and the like. Inclusion of the security sector in the PFM system is essential to achieve the principles of honesty and predictability, i.e. basing the budget on unbiased projections of revenue and expenditure, and stability in policies and funding.

Allocative Efficiency
Allocating fiscal resources efficiently across and within sectors is arguably the most difficult of the three levels of effective public finance. Section 2 has made the case that security counts. Although spending on defense and other security components can enhance growth prospects, given potential inefficiencies in the sector and competing needs from many other sectors, there is a need for (i) a few benchmarks to guide allocations, and (ii) a solid process for making allocation decisions. We focus here on the first element, coming back to the second in Section 5. A first guidepost is to take into account the allocation of fiscal resources in other countries. This is inevitably a simplistic approach given that different countries face very different security-related risks (including due to the differential influence of geography), while costs also are country-specific (given significant economies of scale in the security sector and frequent use of alliances to share costs). In addition, cross-country comparisons are all the more difficult in the security sector because the quality of available data is poor. 8 Finally, this approach also takes a narrow focus on a small number of security institutions, while Section 2 suggested that the perspective needs to be broadened. Nevertheless, and keeping these caveats in mind, based on available data defense spending ranges in most countries between 1-5% of GDP (Treverton, 2005; Table 3).

Source: IMF (GFS), World Bank (2005).
A second guidepost is to assess the rationale for using public resources in the sector. A long tradition of welfare economics has discussed the role of the public sector in implementing various policies. The general lesson is that public expenditures are justified as a correction for market failures (such as externalities, public goods, imperfect information) or for redistribution. Nevertheless, as noted by Devarajan et al (1996), "neither economic theory nor empirical evidence provides clear-cut answers to how the composition of expenditure affects economic growth." On this basis, Paternostro, Rajaram, and Tiongson (2005) provide a framework to assess the issue of allocative efficiency, in three steps. The first question is to assess whether public intervention-as opposed to private intervention-is necessary. The discussion above has shown that dealing with securityrelated risks is often a public good. Defense against external security threats is essentially a pure public good at the national level. Internal public security, supported by the police and justice system, comes close to being a pure local public good for the locality concerned (although access may well be inequitable). Justice is also a public good in that its existence benefits everyone. However, any particular justice transaction benefits (or harms) individuals, and moreover, in many countries people in need of recourse may have a considerable degree of choice as to whether to go through the formal justice system or instead rely on traditional informal mechanisms for this purpose. Landmine clearance has elements of both a public and a private service; the private benefits are high if the land demined is under private ownership, whereas the public benefits are high if it is publicly owned/used land (e.g. roads or irrigation canals). Border management and intelligence are also essentially national public goods. 9 Nevertheless, all these functions might not necessarily be provided by the public sector. For instance, effective rule of law is a public good as it allows enforcement of contracts. However, Dixit (2004) provides a number of examples where public provision might not be needed to ensure contract enforcement, for instance if contract enforcement is based on relations instead of rule of law or where there are private incentives for the enforcement of contracts (this strengthens the point made above about analyzing the investment climate to assess, in this case, what is the constraint on economic growth and contract enforcement). Similarly, arguments about public goods or difficulties in drafting contracts usually do not justify ownership of state-owned enterprises and production of merchandise by the defense sector. This often happens, however; for example, the military in Indonesia raises money outside the government budget through a variety of businesses; these businesses provide paid services; and many of them are not even controlled by the military's central command (Misol, 2006). Beyond the fact that it is not the best use of public resources, this practice raises a number of efficiency problems (badly managed firms) and governance issues (for example poor procurement practices in the sector; refusal of these defense-related firms to pay tax). Second, public intervention does not necessarily mean public expenditure. Other forms of public interventions include regulations and tax policy. 10 Advanced countries, in particular, very often use regulations to reduce security-related risks. An interesting policy question in this regard is the mix of enforcement and fines. Indeed, Polinsky and Shavell (2005) suggest that in general enforcement could be reduced and fines raised, leading to a similar but less expensive level of deterrence. In Section 4, we return to the question of public provision of services (as opposed to public financing of private provision) Third, security-related outcomes depend on various sector interventions, not only those in the security sector. Such synergies between the various sectors need to be taken into account. Not only do these sectors compete for a limited pool of resources, but in addition the key outcomes (such as growth and security) are determined by a combination of expenditures in the various sectors (plus the role of the private sector). The importance of this point was brought out very clearly in our discussion of the risk management framework in Section 2 (with both security sector-related issues influencing allocations in other sectors and the other way around). The three levels of risk reduction, mitigation, and coping indeed suggest broadening beyond the traditional approach to security, which has often been focused on (i) reducing risks, largely through deterrence, by increasing force size, patrolling, etc. This broadening would involve approaches to (ii) mitigating risks (investing in a diversified portfolio of income and assets, which requires a sound growth and political normalization strategy; possibly investment in insurance schemes, either by the formal financial system or informal community-based systems), and (iii) coping with risks (capacity to investigate and prosecute crimes; support to victims). These linkages are important within the security sector as well, for instance in terms of complementarities between prosecution, justice, and the prison system, as is widely recognized in the literature on security sector reform. Security can also be linked with other social risk management mechanisms, such as ensuring macroeconomic stability, developing a sound financial sector, etc. (see also Ghani and Lockhart, 2006).

Operational Efficiency
The third level of this framework consists of ensuring that the public resources appropriated, both for the security sector and for other sectors, are used efficiently and effectively. As in the case of allocative efficiency, a critical requirement is to utilize a range of instruments and processes in the budget execution sphere (see Section 5). In this section, we focus on analytical tools to assess the operational efficiency of security sector spending. Analysis in other sectors typically proceeds with the measurement of unit costs, outcome (or output or effectiveness) indicators, and rates of return (cost / benefit analysis could be used to guide allocations especially within sectors-see World Bank, 2005, Volume 1, Chapter 5-although such analysis has limitations). In theory, the security sector should be subjected to the same discipline. Examples include: • For defense, preparedness indicators (although more akin to output indicators rather than to outcomes) are a highly appropriate performance indicator in peacetime. • For the police, and for the internal security system more generally, the ultimate outcome of local security and rule of law is difficult to measure, especially in Afghanistan where crime statistics are virtually non-existent. Information from surveys on the numbers and types of crime encountered by households could be useful, however. The availability and readiness of police staff, including female police officers and separate detention facilities for women, can be useful indicators, in the case of women officers mitigating the risks women face during detention. • The performance of the justice system can be measured by various indicators like numbers of cases handled, timeliness of processing court cases, and households' perceptions of the justice system and their indications of willingness (or otherwise) to use formal justice channels when seeking redress. • In the case of mine action, outcomes of mine clearance activities (land and areas free of landmines-with clear linkage to ultimate development benefits) can be specified and measured reasonably well, and compared with the costs of mine clearance. • For counter-narcotics, the appropriate outcome indicator at the national level is progressive reduction over time and eventual elimination of the opium economy. However, measuring the performance of counter-narcotics activities is difficult, and linking such performance to progress toward achieving the strategic objective of reducing opium production and trade is even more difficult. Given the variety of activities encompassed under the counter-narcotics rubric, performance indicators for the different components need to be differentiated appropriately. For example, indicators of rural development and agricultural progress can measure the ability of the rural economy to reduce dependence on illicit narcotics production over the longer term. • More generally, in all parts of the security sector some basic indicators of inputs can be useful in measuring performance, for example number and ratio of staff on the job (confirmed by spot-inspections), percentage of staff in security services properly equipped, proportion of staff who have received training, percentage of budget resources spent, etc. Expenditure tracking surveys can be a useful tool in this regard. • Unit costs, although not an outcome indicator, can be useful performance indicators of the efficiency with which inputs for service delivery are being provided. Nevertheless, it must be recognized that this approach has limitations in the security sector, for two main reasons. First, in many cases the objective is to save human lives. Although not insurmountable-the concept of Disability Adjusted Life Year, which provides a common measure for comparing the burden of disease taking into account both death and disability arising from illness, is used in the health sector-measuring the cost of a life is difficult. Second, measuring benefits in the security sector involves low-risk (yet high-cost, even catastrophic) events and numerous prevention activities, also difficult to quantify. It is therefore very important to place security sector institutions in an adequate accountability framework and to follow clear and transparent processes, as discussed in the next two sections.

Security as a Public Service
As noted earlier, security is an essential service that needs to be delivered. Thus taking a service delivery approach makes good sense, and provides some guidance for improving public finance management in the security sector, particularly with respect to institutions, incentives, and accountability. The distinctive features of security as a public service need to be taken into account, however.

Conceptual Framework
A simple model illustrates the institutions, accountability relationships, and financial flows associated with service delivery and their implications (World Bank, 2003b). 11 There are three main actors: the people (as citizens and consumers/clients of services), the state (as political body and policymaker), and the service providers (which could be public or private). The interactions by which these three actors influence and are accountable to each other form an "accountability triangle" (see Figure 3a), and there are correspondingly financial flows among them (Figure 3b). The people can affect service delivery by influencing policy makers (e.g. by voting or through the advocacy role of civil society), who in turn exert influence on the service providers (this is the so-called "long route" of accountability). Citizens can also directly influence service providers (the "short route" of accountability) by selecting the provider (when there is competition), using their voice (e.g. complaints), or making financial contributions (when there is cost recovery). Another important set of actors in developing countries is the donor community, which exerts influence through its dialogue with policymakers, its financing, and its direct contracts with service providers.  One of the main messages emerging from this conceptual framework is that it is often possible to improve service delivery by strengthening the short route of accountability, i.e. through relationships and accountability mechanisms directly between the service providers and the beneficiaries of the service (and their communities). Another key message is the need for donors to reinforce the accountability of service providers rather than inadvertently undermining accountability by making direct payments to them (in the form of cash, salary top-ups, in-kind contributions, etc.). This framework can also shed light on financing issues related to service delivery, specifically that there is scope for cost recovery (a financial flow from beneficiary to service provider) to be an instrument of accountability as well as for covering the cost of services.

Distinctive Features of Security
In applying this conceptual framework to the security sector, it needs to be kept in mind that security as a public service has some distinctive characteristics that affect the way it is delivered and the options for accountability and financing arrangements. First, as discussed earlier, a state of security of the public from violence of all kinds comes close to being a pure public good, meaning that everyone in the locality (in the case of internal security) or in the country (in the case of external security, i.e. defense) benefits from a situation with a general level of security. This implies that the set of actual or potential beneficiaries is no different from the population of the locality or the country, and that cost recovery from beneficiaries is not possible. However, access to public security services (ranging from the police to courts) may be far from equitable for all, and moreover individuals may have very different capacities to mitigate and cope with security-related risks (for instance, vulnerability and coping strategies depend on income and assets: richer households are able to augment their own personal security through expending resources on private security guards, better physical protection of their property, locks and alarm systems, etc.; see Box 1). Second, security in a fundamental sense involves the threat and use of force against internal or external threats to security. This capacity to use force can be misused, as seen in many countries affected by civil conflict as well as in other countries characterized by repressive or predatory states. Even potential-as opposed to actual-misuse of force can generate risk mitigating and risk coping measures that negatively affect people's lives and businesses' prospects. Security forces easily can become a source of insecurity rather than security for the populace. Moreover, monitoring the performance of security service providers can be difficult, as security forces (such as police) are armed and potentially can threaten civilian monitors (see World Bank, 2003b, Box 3.3, p. 54, for a discussion on creating conditions for accountability in the police). Third, and related to the other two characteristics, effective provision of public security involves a monopoly in the provision of the service, i.e. the legitimate exercise of force, which is a fundamental attribute of a functioning state (see World Bank, 2005a, Box 4.1, p. 45). Competition in the realm of security results in insecurity rather than improved security. 12 Fourth, the function of reducing or mitigating security-related risks is in essence difficult to monitor. Some exactions of armed forces for instance are difficult to monitor, because of the threat of use of force or because they happen in remote areas (e.g. border areas).
Preparedness during peace time is also difficult to evaluate. Even monitoring the justice system-the arbitration system of last resort-can be difficult as there may be little recourse to complain. In addition, measurement of outcomes and performance in terms of security is not easy, as the presence of security is essentially the absence of insecurity and associated adverse outcomes. In particular, it would be easy to overlook the implications of security risks in terms of risk-mitigating and risk-coping responses. This issue was discussed in section 3.3 on operational efficiency. Finally, contracting security-related tasks is difficult, notably because outcomes are difficult to monitor, but also because of the issue of use of force. For instance, Hart, Shleifer, and Vishny (1997) review the evidence for and against privatizing prison management: while justice is a public good, one could indeed argue that keeping prisoners is a public good that can be contracted out (a key outcome-prisoners do not escape-is easy to monitor). But the authors argue against privatization, in particular for two reasons: first because of the difficulty of drafting a contract that would ensure good quality (including safety) at a reasonable cost and, second, because to reduce costs, a private prison might recruit under-trained guards who might overuse force to control prisoners.

Implications of a Service Delivery Approach for the Security Sector
As discussed above, the first implication is that security-related risks most often need to be addressed by public intervention. In most cases (but not all-see the earlier discussion on Dixit, 2004), for reasons discussed above (performance oversight, contracting) this should translate into public provision of these services (as opposed to privatization or contracting out). But there are many challenges facing efforts to achieve effective service delivery in the security sector. The first extremely important implication is that effective oversight by the civil authorities is essential. Particularly in the case of the army and other national security forces, the long route of accountability is the only viable one, and this requires that service providers (security forces) are accountable to the state, which is in turn accountable to the population for ensuring the provision of security. Legitimate oversight by the civil authorities with respect to both national security forces and local police is also important because of the risk that otherwise power would be abused. Second, in the case of local security forces providing a localized public good, there is scope to enhance the accountability of the service providers directly to the local population they are serving-while recognizing the risk of elite capture at the local level. Police is a good example. The importance of good police-community relations, which strengthens the short route of accountability, has been widely recognized. More generally, we have seen that a number of the risk mitigation and risk coping strategies are decided and implemented at the community level. In this context, effective complaints channels and other short-route accountability mechanisms are very important, and women and other vulnerable groups in particular need to have access to them. 13 Third, as far as local rule of law and security services are concerned (police, and much of the judiciary), it makes sense to consider some form of de-concentration of government oversight even in an overall unitary state structure, as long as the state is reasonably functional at the local level and not "captured" by regional or local powerbrokers. Fourth, when the short route of accountability cannot be strengthened and the long route of accountability proves counterproductive, an option is independence of some securityrelated functions. A degree of independence is often advisable for the justice system, to insulate it from daily politics. Maskin and Tirole (2004) demonstrate the desirability of "judges" as opposed to "politicians" (in their terminology), contrasting "unaccountable" judges with politicians that could overreact to what they perceive as people's demands and could be insufficiently attentive to minorities. Fifth, despite major difficulties, as emphasized earlier it is critically important to develop monitoring frameworks and performance indicators so that oversight bodies, and ultimately the people, can assess performance with regard to security. Finally, in the case of aid-dependent countries, the importance of the international community reinforcing and not inadvertently distorting or undermining accountability relationships is at least as great in the security sector as in other sectors (Box 2). This implies that international partners should as much as possible work through government budget channels and not build up parallel structures, avoid undermining service providers' accountability relationship with the state (which can happen for example if they provide resources, e.g. in kind or through salary top-ups, directly to the providers), resist "flagging" projects they support to the detriment of the government's visibility, and encourage cost containment for example by supporting competitive procurement.

Box 2: Aid Effectiveness in the Security Sector
As in the case of other themes discussed in this paper, the recommendations for the international community with respect to aid effectiveness more generally (including harmonization, country ownership, use of country systems, etc.) should apply in the security sector. There are however obvious geo-strategic and sovereignty issues to take into account.
In the public finance management area, there is a need to integrate donor funding into the budget so as to maximize the benefits from (i) assessing trade-offs across budget lines; (ii) analysis and evaluation of expenditures; and (iii) ensuring accountability and transparency in use of budget resources. Perhaps the most pressing question is that of fiscal sustainability. As the example of Afghanistan shows, the role of the international community in helping ensure that the security sector is affordable is critical-given its role in making decisions on pay levels and troop levels, its significant financial contribution to the sector; etc. There are also a number of institutional issues to consider in making aid to the security sector more effective, including: (i) fragmentation across security sector institutions (possibly compounded by fragmentation of external assistance if different donors are supporting different institutions); (ii) sequencing of interventions in the security sector (coordinating the strengthening of justice and police for instance given the complementarities of their outputs in relation to the outcome of better security); and (iii) wage practices (including pay levels suggested for troops; top-ups paid to troops; salary levels of recruits in private security firms operating for the international community; etc.). Finally, while this paper has mainly focused on the perspective of developing countries, the international community can also consider trade-offs in its assistance toward security objectives. The cost-effectiveness of external military intervention can be assessed against the cost effectiveness of supporting the strengthening of the local army. Similarly, the cost effectiveness of interventions in the security sector can be assessed against that of other interventions that can be expected to reduce the risks or costs of conflict. For example, increased development assistance, if it results in development progress in the recipient country, could by reducing the risk of conflict there result in lower costs for peacekeeping and future aid. It could also lead to savings on counterterrorism costs in the donor country (see Treverton and Klitgaard, 2005).

Security as a Governance Issue
In this section, we review the security sector from a governance point of view. This is a very important approach since, just as good governance is important in the security sector (as discussed in this section), security itself is important for good overall governance (as discussed earlier). We define good governance to include many dimensions ( Figure 4). As in the case of other aspects reviewed in this paper, good governance principles can and should be applied to the security sector, with appropriate modifications to take into account confidentiality considerations where necessary (but not in a blanket manner or in a way that undermines the fundamental principle of accountability to elected civil authorities). We first focus on public financial management and then review other aspects.

Public Financial Management Processes
Public Financial Management (PFM) processes comprise a core dimension of governance. The role of these processes is to enable strong progress toward the broad budgetary outcomes discussed in Section 3. To analyze the performance of the PFM system, the Public Expenditure and Financial Accountability (PEFA) multi-agency partnership program has identified six critical dimensions of a well-functioning PFM system (PEFA, 2005). Four of these comprise important parts of the budget cycle (policy-based budgeting; predictability and control in budget execution; accounting, recording, and reporting; and external scrutiny and audit-see Figure 5). Another relates to key cross-cutting features of the budget process, i.e. comprehensiveness and transparency, and the last covers out-turns of the PFM system, namely the credibility of the budget. Figure 5 also signals that, given the importance of external financial support in many developing countries, donor practices have a significant impact on the performance of the PFM system (Box 2). We now turn to these six dimensions of the PFM framework. The key message, again, is that good PFM practices should by and large be applied to the security sector as well. The first dimension, PFM out-turns, is very much related to the issues of fiscal sustainability discussed above. Another important aspect is the gap between budgets approved by the executive and legislature on the one hand and actual expenditures on the other. Systematic deviations, often with security institutions spending much more (at least as a share of the total budget if not in absolute terms) than other sectors, are a sign of poor or deceptive budgeting. Such deviations reduce the credibility of the budget, hence weakening its role as a policy tool. On the contrary, even where security requirements change unexpectedly, the resource implications should be managed through the budget process. The second dimension, comprehensiveness and transparency, is of particular relevance to the security sector. Comprehensiveness means that the security sector is fully incorporated in the budget and budget process. Comprehensiveness is sometimes challenged by real or alleged confidentiality issues (see next paragraph). Another challenge is the role of the international community, which often supports the sector financially but without using normal budget systems (see Box 2). Comprehensiveness also means that all sectors (including security) must compete on a level playing field for funding during budget planning and formulation. This does not imply a mechanical or "equity" based allocation of resources, but rather that different sectors contest for resources during the budget formulation process on an equal footing based on their role in the national strategy and (for sub-sectors) within the corresponding sector strategy. 14 Transparency in the security sector is contentious given the nature of the sector, which in some specific cases requires confidentiality leading to a culture of secrecy. Nevertheless, traditional arguments in favor of transparency apply to the security sector as well: (i) sharing information allows for better political participation and better policy decisions; (ii) it fosters coordination with other sectors (security has to be seen in a broad context); (iii) monitoring is important to ensure accountability and fight bureaucratic inertia; (iv) transparency encourages self-restraint (potentially desirable to avoid risk-mitigating reactions); and (v) overall, transparency supports better implementation (see Roberts, 2007). This is not to deny that there are areas where secrecy is desirable, for instance to protect judicial or intelligence sources or to create a strategic advantage of surprise vis-à-vis an enemy. But there is also the risk that secrecy directives would be excessive and carry consequences (according to Roberts, 2007, the ordinance on state secrets in Vietnam is so broad that some economic data generally available in other countries are classified). Hence security sectorspecific exceptions to transparency should be limited and clearly understood. In particular, there is little or no justification for withholding general information on agency budgets (including budgets of intelligence agencies). The third dimension, policy-based budgeting, covers processes by which resources are allocated according to strategic priorities both within the security sector and across all sectors (including security). Sector strategies are critical in this regard, as are the broad priorities indicated in the overall national strategy for development. Figure 6 presents a generic process for defense policy and organizational reviews, which is broadly applicable to other parts of a country's security sector as well. While the full complexity shown need not be followed in every case, the main elements need to be present.

Force Structure Options within Context of Financial Parameters
Including force structure, materiel, end strength, procurement, infrastructure, readiness

Strategic Security Assessment
Domestic, regional, international environment National commitments Potential risks/challenges

Non-defence Security Policy Process
Intelligence; public security; intelligence, and so on

Guidance on Review Process
Overall policy direction Issues to be addressed Fiscal framework Required consultations Due date

Economic Policy Framework
Including national development objectives, defense budgeting process
These policy development and organizational review processes need to be guided from the beginning by an indicative fiscal framework to promote sustainability (see Section 3 on fiscal affordability). Review processes should be as transparent and inclusive as possible. Insofar as they are based on broad consultations among the relevant stakeholders, and if the final product is made public, for example through the publication of a policy paper, their legitimacy will be strengthened. Of key importance here is the economic framework, and its linkage with setting the defense policy framework (in the case of defense, the strategy needs to be based on a transparent "threat assessment" fully owned by the government), assessing different fiscal options, and ensuring oversight by relevant executive and legislative bodies. Finally, prioritization must take into account issues of resource allocation over time. A classic misallocation of funds found in many countries is building physical assets like infrastructure, or service delivery networks like education and health systems, without planning for or even taking into account the associated recurrent expenditure requirements. The same can occur in the security sector, where lethal and non-lethal equipment may be procured, facilities constructed, or security forces expanded, without consideration of their full life-cycle costs. While in annual budgets this will be manifested as an imbalance between capital and recurrent expenditures, in particular shortfalls in non-wage recurrent spending, from a medium-term perspective the problem is one of prioritizing between present and future expenditures in line with sustainable force sizes and administrative and service delivery structures. Thus a medium-term fiscal perspective is an essential tool for prioritization as well as for assessing overall fiscal affordability (see above). The fourth dimension, predictability and control in budget execution, covers various areas ranging from internal control (payroll and non-salary) to procurement. 15 The control framework, and also the accounting categories and reporting process (see next dimension), are very important for effective and efficient utilization of budgetary resources. Hence good practice calls for standardizing these important components of the PFM system in all sectors, including the security sector. One area where different procedures are followed in many countries in the security sector, especially in the case of defense, is procurement. The overriding objective of a national public procurement system is to deliver efficiency and value for money in the use of public funds, while adhering to fundamental principles of non-discrimination, equal treatment, and transparency. Procurement is therefore at the core of the PFM system and contributes greatly to several of its objectives, including efficiency, transparency, and accountability. Defense, security, and budgeting specialists generally take the position that security sector procurement and acquisition practices should adhere to the same principles that guide procurement in non-security parts of the public sector: fairness, impartiality, transparency, cost-effectiveness/efficiency, and openness to competition. In addition, it is essential that there be systematic evaluation of all major procurement transactions for all forms of public sector procurement. However, defense budgeting specialists have also noted that-with the exception of procurement of works and most commodities (such as construction, clothing, food, fuel, office equipment, general vehicles, and consultancy services)-defense acquisition in particular (and also to a lesser extent other security acquisitions) does exhibit some distinctive characteristics. These relate to: (i) the relative importance of cost in determining which bid is accepted; (ii) confidentiality related to national security considerations; (iii) the time frame required for major weapons procurements; (iv) complexity of defense procurements; and (v) the existence of international arms control treaty regimes and national legislation governing arms procurement. These distinctive characteristics are deviations in application rather than of principle. There certainly should be skepticism about any claims that procurement of relatively standard works, services, and commodities for defense or any other part of the security sector should be subject to different rules. More specifically, considerations of confidentiality need not and should not prevent the key principles of good procurement practice from being fully applied throughout the security sector. Not doing so runs the risk of limiting competition, raising costs, and exacerbating vulnerability to corruption (which can be a significant risk particularly in the case of larger defense procurements). While as noted, confidentiality is a legitimate consideration for certain types of procurement in the security sector (e.g. advanced defense equipment), much of the procurement for security forces (food, uniforms, transport vehicles, small arms and ammunition, etc.) does not involve overriding confidentiality considerations. In any case, it is important that sound public procurement principles and procedures be applied to the security sector, with appropriate provisions to ensure confidentiality where necessary. The fifth dimension, accounting, recording, and reporting, covers a number of key financial management processes. These are important because they serve the objectives of transparency, accountability, and performance evaluation. In line with the principles of comprehensiveness and consistency of treatment of different sectors, the security sector should be covered by the government-wide treasury system-otherwise financial information for the security sector will not be available on a comparable basis with other sectors throughout the budget cycle. Related to this, classification systems used for security (including defense) need to be fully consistent with those used in the rest of the budget. The sixth dimension, external scrutiny and audit, covers external audit and oversight, the need for which (with appropriate provisions in line with confidentiality concerns as necessary) is just as great in the security sector as in other sectors. The Auditor General should have sufficient access to transactions and performance indicators in the security sector, and its audit reports should be adequately discussed in Parliament (which should also be allowed to engage in meaningful debate over the budgets of security sector agencies). If needed, and as discussed above, this process (both with the Auditor General and in Parliament) could be restricted to smaller groups that have adequate security clearance (e.g. through a special Security Committee in Parliament). Treverton and Klitgaard (2005) suggest a number of questions to raise about the quality of external scrutiny: (i) Are there clearly defined executive and legislative responsibilities for external and internal/domestic security?
(ii) Are the security forces subject to democratic control? (iii) Are parliamentarians, the media, and civil society free and capable to participate in the security debate and are they doing so? (iv) Are the security forces able to exercise political influence? (v) Are the security services open to unnecessary political interference through political reach-down into the promotion system? (vi) Are the security forces more loyal to the regime or to the people? (vii) Are there budgetary checks, balances, and internal and external audit, and are these transparent? (viii) Are the duties and responsibilities of the security services enshrined in legal statutes, military law, and codes of conduct?

Other Aspects of Governance
The security sector is particularly vulnerable to corruption, not least because of its culture of secrecy which leads to limited transparency and reluctance to allow civilian and external oversight. The difficulties of measuring and assessing performance further increase vulnerabilities to corruption. Bribes can dilute deterrence against violations of law (Polinsky and Shavell, 2005). Corruption is all the more problematic in the security sector because the sector typically is expected to play a major role (of deterrence and prosecution) in the fight against corruption. In addition to sound public financial management, other aspects of a good governance system include (see Figure 4): 16 • Political oversight and accountability: including by Parliament (discussed above).
• Civil society oversight (see also the previous section): it is important to give sufficient access to information (on policy choices and on actual implementation) to various groups in civil society (media, groups of experts in academia, private sector, other members of the public). • Interface with the private sector: whenever possible, the private sector should be given a role in oversight and providing checks with respect to the security sector. For instance, the private sector can be appropriately involved in oversight of public procurement for the security sector to ensure good governance and value for money. Justice and police are a key driver of the investment climate, and the private sector should have opportunities to voice its concerns in a constructive manner. • Public sector management: in addition to public financial management (see previous section), good public sector management requires several elements. First is civil oversight in the security sector ministries (as well as, in the area of PFM, in the Ministry of Finance) and through the President's Office or National Security Council. This has to be complemented by strong, effective, and ethical leadership (which itself can be complemented by some technical measures, including asset declaration, ethical norms or codes of conduct for personnel in the security sector, etc.). Second, personnel management is critical for the effectiveness and legitimacy of the sector. Experience in the civil service underlines the importance of merit-based processes, ethnic diversity, and avoiding patronage and its adverse impact. An important issue, which relates to corruption, is the salary level of personnel in the armed forces and in the judiciary, which needs to be sufficiently attractive while remaining fiscally affordable and not distorting the labor market. Third, good public sector management requires effective coordination among agencies. This is often linked to ethnic issues, particularly if various institutions are dominated by different ethnic or political groups, as in the case of Afghanistan, Palestine, and often Lebanon. • Decentralization and local participation (this was discussed in Section 4, showing some potential but important limits to this factor for the security sector). • Cutting across these issues is the role of the international community, reviewed in Box 2.

An Application to Afghanistan 17
The Afghanistan Public Finance Management Review included a substantial analysis of public expenditures and their management in the security sector (see World Bank, 2005b, Volume V). This work, carried out jointly with the UK Department for International Development, responded to a request from Afghanistan's Ministry of Finance to include the security sector within the scope of the Review. It provides an illustration of how many of the themes, principles, and methods discussed earlier in this paper can be applied to analysis as well as institutional and policy recommendations for improving public financial management in the security sector of a specific conflict-affected country. Nearly a quarter-century of protracted conflict in Afghanistan, starting in the late 1970s, led to pervasive insecurity and destroyed or profoundly damaged the institutions that had been responsible for providing security in the pre-war period-both formal governmental institutions and traditional mechanisms. Government security forces were misused in ways that reduced rather than enhanced security, e.g. during the Soviet occupation of the 1980s, and they became increasingly "captured" by factions in Afghanistan's civil war during the 1990s. The emergence of numerous well-armed militia groups, in shifting loose alliances, meant that the ability to use large-scale violence in pursuit of objectives was no longer the monopoly of the government but rather was widely contested, greatly undermining security (see Section 2 and Figure 2). The judicial system was similarly undermined, and traditional mechanisms for providing justice were often captured and used for their own ends by commanders and warlords. Women suffered greatly from the conflict as a result of violence against them, and also due to limited mobility and lack of access to education and health services, culminating in the Taliban's severe social restrictions. Neighboring countries and other powers repeatedly and competitively intervened militarily-either directly or indirectly by providing arms and other support to different factions-contributing in a major way to insecurity in Afghanistan. In sum, the long period of conflict and insecurity had profoundly damaging economic, social, and political consequences for the country and its people. The critical importance of enhancing security and reforming the security sector in Afghanistan has been widely recognized. This was signaled in the Bonn Agreement, which provided for the establishment of an International Security Assistance Force (ISAF), and two subsequent donor meetings on security set targets and assigned donor responsibilities.
Significant progress was achieved in the initial years following 2001 in strengthening and in certain respects reforming the security sector. Successes included most generally the steady shift away from a pattern of endemic factional and ethnic conflict, the gradual extension of the writ of the national government beyond Kabul, the formation and gradual build-up of the Afghan National Army, and implementation of the DDR process. However, great gaps remained both in terms of areas that the Bonn Agreement did not cover and in achieving agreed security-related targets. Moreover, international security forces' presence in Afghanistan was initially very modest and limited as compared to other post-conflict situations. In 2005 and especially during 2006 the security situation deteriorated with the resurgence of the Taliban insurgency in the south. This deterioration has continued in 2007 and 2008, with the insurgency spreading to new areas.

Salient Features and Key Issues
Salient features of Afghanistan's security sector at present include the following: • The historical legacy of conflict was fragmentation into regional and local militias, capture of policing and military functions by non-legitimate actors, and erosion of both formal and traditional justice systems, also with capture by non-legitimate actors. Building the state since late 2001 has entailed reconstituting and reforming the security forces under legitimate oversight by the civil authorities. • The overall size of Afghanistan's security sector, even with recent and planned expansion, does not seem unreasonably large by international standards. The planned size of the ANA was set at 70,000, subsequently increased to 80,000, and most recently a sharp further increase to 122,000 (plus 12,000 in training pool) has been targeted. The planned size of the Afghan National Police (ANP) was 62,000, later augmented by some 11,000 auxiliary police, and subsequently expanded to an overall ANP force ceiling of 82,000, with a further increase under consideration.
• Nevertheless the security sector is costly and most likely unaffordable. Over the medium term, it is extremely doubtful whether the sustaining costs of Afghanistan's security sector at planned force levels can be absorbed by Afghanistan's national budget. This was an important finding in World Bank (2005b, Volume V), and subsequently the fiscal cost of the security sector has further escalated due to substantial increases in force sizes as well as salary increases and growth of other costs.
• Progress in developing different parts of the security sector has been uneven.
After a slow start initially, the formation and expansion of the ANA have moved forward rapidly. Progress in developing and improving the ANP has been much slower, and reform and capacity building in the justice sector have lagged far behind.
• Security and fiscal issues are closely interlinked (see Section 3.1). For example, low domestic revenue is both a consequence and a cause of insecurity. Costs of reconstruction tend to be higher due to insecurity, while the security sector is a source of considerable fiscal pressure. Pay and grading decisions with respect to the security services, made outside national budget processes, have important direct fiscal implications and secondary consequences through pressures on civilian pay. Trends in security sector expenditures, including both Core Budget (spending included in budget documents and going through Treasury channels) and External Budget (spending directly executed by donors and their contractors), illustrate some of these points (Table 4). From the perspective of this paper, key challenges facing Afghanistan's security sector include the following: • The role of policy-based budgeting (see Section 5.1; these points are reviewed in more detail in Section 6.3). First, an integrated security sector strategy and policy framework is needed; otherwise sound programmatic-e.g. on force sizes, equipment-and public expenditure decisions cannot be made. So far the National Security Strategy has not yet been finalized, although reportedly its preparation is at an advanced stage. Second, very high donor-executed spending through the external budget renders achieving coherence of spending across and within subsectors much more difficult, and also raises questions about the fiscal sustainability of the security sector in the medium term. Third, there are many difficulties in coordinating and prioritizing security sector expenditures and actions, reflecting the above problems and fragmentation of decision-making across donors. These factors make it more difficult for the National Security Council (NSC) to fully carry out its mandated strategic, leadership, and coordination functions, although the situation in this regard improved in 2004 and 2005.
• Concerns about the growing security sector wage bill and pressures for fiscally unaffordable salary increases, as well as the fiscal sustainability of security sector staffing levels. The ANA salary structure, determined without reference to fiscal constraints or pay in the civil service, set a fiscally costly precedent which also resulted in a large pay increase for the police, and which other sectors aspired to (this point is discussed in more detail in the next section). Subsequent further increases in ANA salaries have exacerbated such pressures and tendencies.
• The role of good governance (discussed in Section 5.2). Public administration reforms and capacity building in security sector management and oversight institutions have been generally slow, lagging behind the development of security forces like the ANA. This imbalance, as well as the lack of progress in some security subsectors (e.g. justice), carries serious risks. It is essential to develop good governance and sound and sustainable financial management practices throughout the security sector, in line with and integrated with national budget processes and fiduciary provisions.

Toward a Fiscally Sustainable Security Sector
Decisions on and investments in force levels, salaries, non-salary staff-related costs, standards of equipment, base construction, etc. in the security sector are leading to a fiscally unsustainable situation-for the security sector itself and for Afghanistan's budget as a whole. Figure 7 below illustrates the dilemma that for some years to come, planned/forecast levels of security sector expenditures will exceed likely domestic revenues, under a somewhat ambitious yet reasonable "base case" of revenue mobilization and even more under a "low case". These expenditure forecasts are conservative and based on earlier targeted ANA and ANP sizes, and do not include any provision for further rises in wages or increases in other costs. And in any case, allocating the bulk of domestic revenues to the security sector is not viable or desirable over the medium term. This situation, which stems from the large (albeit of uncertain duration) availability of external resources outside the Government Core Budget (i.e. in the External Budget), fragmented decision-making and lack of cost-consciousness, difficulties faced by the Government in exercising strategic coordination of security sector spending, and limitations on oversight by entities like the Ministry of Finance, urgently needs to be addressed. A recommended set of near-term actions includes the following: • Greatly enhance the information base on security sector expenditures, especially the cost implications of planned expansion of security forces, and also of associated wage structures and non-wage recurrent expenditure requirements.
• Review the fiscal implications of ongoing security sector expansion and reforms. The Government should conduct an integrated security sector-wide fiscal review to forecast the fiscal implications of currently programmed security sector expansion and reforms over the medium term. The review could be done under the overall umbrella of the Medium-Term Fiscal Framework (MTFF) into which it would feed. The sustainability of planned expenditures needs to be assessed within the context of the overall rolling fiscal plan, growth of domestic revenues, and possible medium-term commitments from donors to fund recurrent costs.
• Revisit strategic objectives and trade-offs in light of fiscal constraints. This could be done as part of finalizing the National Security Strategy (see Section 6.3).
• Take actions to improve the situation. These could include (i) containing further security sector wage increases; (ii) reviewing programmed force sizes, especially for the ANA, in light of fiscal constraints; (iii) exploring possibilities for containing nonsalary recurrent costs while not adversely affecting readiness and operational capabilities; and (iv) similarly reviewing construction plans and equipment costs.
• Any changes in security sector expenditures resulting from unforeseen developments in the security situation and requirements should occur through the budget process (including a revised budget during the financial year if necessary), not in an ad-hoc manner.
• Explore with donors prospects for predictable medium-term external funding for the security sector, in line with agreed security sector objectives and a clear road-map for eventually phasing out such funding over a number of years as Afghanistan progresses toward a sustainable fiscal position and security improves.
• Ensure that the fiscal sustainability "lens" is applied to future decisions on the security sector, with a view to preventing decisions or actions from worsening the fiscal situation of the security sector.

Strategic Prioritization and Sequencing
As can be seen from Figure 8 below, security sector expenditures have been dominated by Defense (primarily ANA) and Interior (mainly police), which together account for close to 90% of the total. Counter-narcotics expenditures have been variable and in part may be double-counted with other spending inside and outside the security sector (hence they are not shown in Figure 8), but they also have become quite large. The low spending on the justice sector is striking and is reflective of the lack of progress until recently on improvements and reforms in the justice system. The national police also was underfunded in the initial years after 2001, although this was subsequently corrected, primarily through large External Budget expenditures. Thus the overall allocation of security sector expenditures seems to have been somewhat disconnected from the concerns voiced by large numbers of Afghans about security and crime. As for the national budget as a whole, there are no easy answers or technical fixes for allocating public spending, either between security and other sectors or within the security sector. Clearly, the very large share of security spending in the External Budget (nearly 80% in 2004/05), and fragmented donor decisionmaking on expenditures, make the task of the National Security Council, the Ministry of Finance, and other oversight bodies especially difficult. Some points with respect to intersectoral and intrasectoral prioritization are outlined below (see World Bank, 2005b, Volume I, Chapter 5 for further discussion).
Prioritization across sectors: Sophisticated technical approaches may require substantial resources and even so are highly imperfect and can result in misleading conclusions. Simpler rules of thumb may work better, and judgment is called for.
• Intersectoral allocation decisions need to be anchored in the overall national security strategy and national development strategy. • Identifying and mitigating gross anomalies in intersectoral allocations is very important (e.g. the serious underfunding of the justice sector, noted earlier). • International patterns of spending and outcomes (for example for police) may provide some guidance on where funds should be allocated (see Section 3.2). • Identifying bottlenecks and directing expenditures to alleviate them may be a useful approach in the short run. • Some linkage between sectoral allocations and performance may make sense but needs to be watched carefully, and corrective actions taken if required, to ensure that large imbalances between sectors do not emerge over time. • Finally, it is essential that intersectoral prioritization be comprehensive-no sector (e.g. Defense) should be considered sacrosanct or immune from scrutiny.
Prioritization within sectors: Technically, prioritization within sectors is more manageable than intersectoral prioritization because the alternatives tend to be more comparable.
• Intrasectoral allocation decisions need to be anchored in a sound sector strategy.
• Technical analysis (e.g. cost-effectiveness analysis) can be used to evaluate individual project proposals and also to make more general prioritization decisions. • Ruthless culling of projects with lower benefits and/or which do not have good prospects for rapid completion will improve the timeliness, efficiency, and effectiveness of sector programs. • Complementarities need to be factored into prioritization decisions (e.g. there is no point in having a police force that is not adequately armed, clothed, equipped, and provided with vehicles and fuel). • Correcting geographical and other disparities (for example in terms of police coverage) may help guide prioritization.

Budget Execution for Effective and Efficient Utilization of Funds
Effective and efficient utilization of funds to achieve stipulated purposes is a key element of the PFM system, including in the security sector. As we have argued, a general principle is that all elements of a sound budget execution process should be applied to the security sector, in the same way as to the rest of the budget. Security expenditures in Afghanistan's Core Budget are essentially subject to similar control mechanisms, accounting and reporting requirements, and procurement arrangements as in the rest of the Core Budget. The main difference is the absence of a strong internal audit function-a role that in the case of the civilian budget is to a large extent played by the Monitoring Agent of the ARTF (see World Bank, 2005b, Volume I, Box 6.2). The absence of such a mechanism-albeit extraordinary and temporary-in the security sector means that there is a weaker control framework despite the application of other elements of the Core Budget execution process. Security spending in the External Budget occurs completely outside these Government processes. This carries serious disadvantages with respect to the quality and consistency of information flows, and also procurement. Another major problem has been the variety of salary levels, payment channels, and top-up arrangements in the security sector, which risk fragmenting the accountability and loyalties of security sector staff. The shift of ANA salary payments and food expenditures for the ANA onto the Core Budget represented a major improvement in this regard. With respect to procurement, it is possible to follow sound procurement procedures despite the actual or perceived obstacles in defense or other security sectors, and good procurement practices will reduce vulnerability to corruption (see Section 5.1). In particular, competitive procurement with adequate transparency and oversight is critical. With the recent passage of a new Procurement Law that meets international standards, donors funding security through the External Budget should move toward using sound internationally acceptable procurement procedures, either those provided for in Afghanistan's Procurement Law or similar procedures. This is also important in the case of procurement of supplies, security services, etc. by international military forces in Afghanistan. There is also a rather formidable agenda for generating and effectively utilizing monitoring information (Section 3.3). This can start with small steps and make progress incrementally over time. One promising option is to generate and regularly monitor basic financial data on on security sector expenditures in the Core Budget, and encourage donor partners to provide similar information on as consistent a basis as possible for External Budget expenditures in the security sector. Staffing data, which have suffered from inconsistencies, need to be improved and monitored. These can be augmented as proposed above with indicators of numbers or proportions of staff on duty and available to carry out their responsibilities. Subsequently, simple indicators of output/performance could be compiled and used in monitoring. It is more important to collect and effectively utilize simple information than to devise highly sophisticated data and monitoring systems which are difficult to implement and become a burden on scarce Government capacity. Finally, the institutions concerned need to take on the role of monitoring and evaluation and treat it as a serious part of their work, and to make meaningful use of performance indicators in decision making.
To illustrate how straightforward monitoring of basic financial data can provide meaningful insights and feedback for resource allocation and other budgetary decision-making, Table 5 compares estimated actual expenditures with budget targets for various categories of security sector expenditures in 2003/04 and 2004/05. Key points that emerge include the following: • There were sizable deviations between budgeted and estimated actual spending for most categories of security expenditures and also major fluctuations over time, suggesting that there is still a long way to go before the budget becomes a reasonably good predictor of actual expenditures. • Within the security sector, External Budget expenditures are substantially higher in relation to budget targets than are Core Budget expenditures. • Defense expenditures tended to perform better in relation to budget targets than other parts of the security sector, especially in 2004/05. • Comparing the security sector with the budget as a whole (see World Bank, 2005b, Volume II, Part 1, Table 1.1), it is clear that the security sector achieved considerably higher expenditures in relation to budget targets than did the overall budget.
appropriately integrated into all aspects of the country's PFM system: medium-term fiscal programming (MTFF), annual budget formulation, staffing and payroll management, financial management and controls, and external scrutiny, among others. There is no justification for treating the security sector as separate or sacrosanct, and not subjecting it to budgetary and fiduciary processes. Specific recommendations include the following.
Review the Fiscal Implications of Ongoing Security Sector Reforms: Enhance the information base on security sector expenditures, and conduct an integrated security sectorwide fiscal review to forecast fiscal implications of security sector development and reforms over the medium-and longer-term. This would serve as a basis for considering security sector force levels, and for discussions with donors on longer-term financing. It would also facilitate policy-and program-based budgeting of security sector expenditures with a medium-term perspective.
Further Develop Policy Making Capacity in the NSC and Finalize the National Security Strategy: The National Security Advisor and NSC are tasked with coordinating policy development and overseeing integrated strategic planning across security institutions. Adequate capacity is required for these purposes. An integrated security sector strategy and policy is needed to determine acceptable longer-term costs of security institutions and appropriately prioritize and sequence spending across and within subsectors (e.g. justice and police versus ANA). The Government of Afghanistan should finalize the National Security Strategy, and it will need to gain wide buy-in within and outside of the Government.

Treat Security as an Integral Part of the National Development Strategy: The recently
completed Afghanistan National Development Strategy can be used as an opportunity to ensure that the security sector both is accorded its due importance as a development issue for Afghanistan and that the broader developmental and fiscal perspective is injected into the security sector strategy.

Implement Administrative Reforms and Develop Security Sector Management and
Oversight Capacity: Reforms are needed to enhance human resource capabilities in the Ministries of Defense, Interior, Justice, Counter Narcotics, and in the National Security Directorate. A particular priority is developing solid financial management capacity and practices in these line ministries, including strengthening the control framework, notably with the development of internal audit.

Rigorously Enforce Staffing Size, and Set Salary Increases Taking into Account Fiscal Absorption Capacity:
It is vital that agreed formal staffing establishment sizes be enforced. Moreover salaries, like non-salary costs, must be subject to the normal rigors of public expenditure management and fiscal discipline-the security sector wage bill must remain affordable.

Role of donors:
Donors play an even greater role in Afghanistan's security sector than they do in other sectors. Consequently they are also important in public financial management in the security sector. Whereas donor practices inadvertently can hinder the Government's efforts to make the national budget the central instrument of policy, reform, and resource allocation for the security sector, donors can also make a major contribution to improvements in this regard. Specifically, donors can contribute to improving public finance management in Afghanistan's security sector by: (i) supporting finalization of an agreed National Security Strategy and corresponding strategies and policies for the individual sectors; (ii) reassessing views about force sizes and cost requirements in the light of fiscal constraints; (iii) helping enhance the capacity of oversight actors (such as the Auditor General, legislature) and economic managers (Ministry of Finance) to address issues relating to financial management in the security sector; (iv) channeling more of their assistance to the security sector through the Core Budget; (v) for resources not channeled through the Core Budget, ensuring that timely and accurate information is made available to the budget authorities, and maximizing use of internationally acceptable procedures e.g. for procurement; (v) helping key security ministries strengthen their budget formulation and budget execution processes; and (vi) discussing medium-and longer-term availability of external resources for the security sector. Implementing these recommendations and other initiatives to improve security in Afghanistan, which is a necessary condition for the country's development, undoubtedly will take time, particularly in view of the intensifying conflict. However, there is a need to get a handle on the fiscal sustainability issues in the security sector, as there is a risk that decisions on force levels, pay, non-salary spending, equipment, construction, etc. that are being made or implemented today will be unaffordable for Afghanistan in the future. Dialogue with donors, including on predictable medium-term financial support to Afghanistan's security sector, also is urgently needed.

Conclusions and Entry Points for a Two-Way Dialogue
Security and development are intimately related, as is recognized in a growing body of literature and increasingly by key development actors. In this paper, we have argued that understanding security-related risks and designing risk mitigation strategies are crucial for achieving better development outcomes in terms of growth, poverty reduction, and welfare in developing countries. To this end, a number of economic tools can be brought to bear to complement traditional approaches to assessment of security issues and risks. Among these tools, we have focused on those related to public finance management, service delivery, and governance, because of the positive experience with these tools in other sectors as well as their applicability to the security sector. A key lesson is that in almost all cases traditional tools and principles can and should be applied to the security sector (with appropriate-but not excessive-scope for adjustments in view of confidentiality considerations). Public finance has important linkages to security (as shown in this paper) and to economic growth and development (as has been widely recognized in the economic literature). The security sector will function better, and in a more sustainable manner, if good PFM principles are applied in the security sector. This linkage also works through the impact of sound public finances on the economy and thereby indirectly on security. On the other hand, better security can also support improvements in public finances, most notably through enabling increased revenue mobilization, which in turn can provide fiscal space to spend more on the security sector, and so on. Breaking out of an "informal equilibrium" where insecurity and weak public finances (as well as poor governance, lack of capacity, etc.) reinforce each other, and moving toward a "virtuous circle" whereby improvements in these areas become mutually reinforcing, is critical in insecure, low-capacity, often conflict-affected countries like Afghanistan. Using economic tools effectively in the analysis of security issues and of the nexus between security, development, and public finances-and for developing and reaching consensus on concrete, practical policy recommendations-will require a two-way dialogue. Ministries of Finance, civil society, bilateral donors, and international financial institutions, which are generally familiar with these tools, should start applying them to the security sector. The benefits of doing so have been demonstrated in the Afghanistan case just discussed. It is important that the application of economic tools not be mechanical but rather contextualized in the light of country circumstances, including based on adequate knowledge of a country's security sector characteristics and its budget system. From the other side, actors directly involved in the security sector should familiarize themselves with these tools and associated PFM, service delivery, and governance themes to facilitate productive dialogue. Such actors can help provide the contextual information and knowledge on the security sector that will enable the application of PFM analysis and other economic tools to be fruitful. Possible entry points for such a dialogue between development/public finance management entities and security sector institutions include the following: • Policy integration. Emphasize that security is a key factor for development, for instance by including an MDG goal dedicated to security (as in Afghanistan) or integrating security in the poverty reduction strategy (see World Bank, 2005c). And view development as a longer-term driver of improved security (see Box 2). Policy integration can be encouraged through analytical work at the country level establishing the importance of these linkages, and by holding development and policy forums with sessions on these subjects that include both sets of actors as participants.
• Fiscal issues. Integrate the security sector in the dialogue and frameworks for fiscal issues-both short-run fiscal management and longer-term fiscal adjustment as well as fiscal sustainability considerations. Fiscal linkages are particularly important to Ministries of Finance, which can be persuaded of the importance of these linkages and to take them up in their dialogue with the line agencies concerned. In donor countries as well, issues of fiscal sustainability are important for Finance Ministries.
• Public finance approach. Bring to bear the key lessons learned in public finance management (on sustainability, allocative efficiency, transparency, etc.) and service delivery (decentralization, accountability, etc.) and their implications for the security sector. Security sector actors are interested in whether and how their expenditure demands can be met, which provides a possible entry point for bringing them into the dialogue.
• Policy and implementation coordination. Develop institutional mechanisms to enhance the integration of the security sector with the rest of the development strategy and national budget (at policy and implementation levels). Appropriate institutional mechanisms are needed to bring together domestic actors (possibly through establishing and strengthening a National Security Council; emphasizing the need for transparency), international actors (coordination between multilaterals-e.g. international financial institutions and the UN system, coordination within bilaterals-development arms and foreign affairs and defense arms), and between them (through consultative groups, Poverty Reduction Strategy mechanisms, etc.).
• Skills. All this requires development of more diverse skills in the various institutions to allow a common language and productive dialogue. Considerable skill development can occur through learning by doing on both sides, which can be accomplished by programming integrated analytical work and joint assessments. This will require some resources, including notably the time of concerned staff. Much more remains to be done in terms of data collection and research, which can fruitfully shed light on the issues. In the meantime, however, it is hoped that these possible entry points, together with the example of this framework applied to the case of Afghanistan, will provide a meaningful basis for (i) initiating a productive dialogue between security and development / public finance actors, (ii) exploring further the lessons from experience, and (iii) together coming up with recommendations for improving policies, processes, and institutions that will make a difference.