WPS6595 Policy Research Working Paper 6595 Behavioral Economics and Public Sector Reform An Accidental Experiment and Lessons from Cameroon Gaël Raballand Anand Rajaram The World Bank Africa Region Poverty Reduction and Economic Management Unit September 2013 Policy Research Working Paper 6595 Abstract Starting with the hypothesis that behaviors are the by a small trade facilitation grant of approximately critical (and often overlooked) factor in public sector US$300,000. One approach appears to have succeeded performance, this paper explores the notion of how in initiating change while the other has signally failed. behavioral change (and thus institutional change) might The two projects of different scale, scope and design be better motivated in the public sector. The basis for in the same governance environment offer a very this study is “an accidental experiment� resulting from interesting natural experiment (unplanned but accidental the World Bank’s operational engagement in Cameroon. for that reason) that allows insights into the nature In 2008, World Bank staff successfully concluded of institutional change and the role of behavior and preparation on a project to support the Government incentives and approaches that offer greater prospects for of Cameroon to improve transparency, efficiency, and making reform possible. The paper confirms the value of accountability of public finance management. The using ideas from behavioral economics, both to design US$15 million project supported a number of ministries institutional reforms and to critically assess the approach to strengthen a broad range of management systems and to institutional reform taken by development agencies capacities. Independently and concurrently, other Bank such as the World Bank. staff initiated a low-profile, technical assistance project to improve performance in Cameroon’s Customs, supported This paper is a product of the Poverty Reduction and Economic Management Unit, Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at arajaram@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Behavioral Economics and Public Sector Reform An Accidental Experiment and Lessons from Cameroon Gaël Raballand Anand Rajaram 1 JEL: F35, H83, K42. Keywords: customs, Cameroon, aid effectiveness, behavioral economics, governance, capacity building. Sector board: PSM. 1 Gaël Raballand and Anand Rajaram are respectively senior public sector specialist and Africa region governance practice leader for the World Bank, e-mail: arajaram@worldbank.org (corresponding author). The authors would like to thank Gregor Binkert, Thomas Cantens, Mamadou Deme, Frédéric Gaspart, Nick Manning, and participants at the March 2013 CSAE annual conference for useful comments. The nature of management reform is such that an assessment cannot be solely grounded on hard evidence of what succeeded or failed. One’s own judgment must be brought to bear, as well as that of others – participants and observers – who have seen the reforms in operation and have thought much of how the system is operating. – Allen Schick (1996) 1. Introduction To paraphrase Milton Friedman, “we are all institutionalists now�. Economists acknowledge that policies are only as good as the institutions which design and implement them, political scientists agree that institutions are the critical element in good governance, and practitioners of development assistance emphasize the need for solid institutional foundations for sustainable development. Yet, the actual task of building and strengthening key public institutions remains a problematic enterprise. How are good institutions established and how are poorly performing ones reformed to meet current needs? What are the strategies and processes that drive reform? What is a realistic timeline for fragile states to put in place a functional public administration? There is a large literature on the broad topic of public administration, or what is now more conventionally referred to as public management, which would be expected to provide some of the answers to these questions. Much of this literature has focused on the challenges of achieving effective and efficient government in the developed countries. Politt and Bouckaert (2000) provide a widely cited review of the state of knowledge based on the experience of public sector reform in ten OECD countries, (Australia, Canada, Finland, France, Germany, New Zealand, the Netherlands, Sweden, UK and the USA). Politt et al. (2000) suggest that public sector reform is typically a top down response by the political system to concerns about fiscal pressures or issues of legitimacy. Using a framework in which the political system is seen to interact in a triangular relationship with the system of public administration and the market economy, they provide a useful description of the various approaches to public management reforms across the 10 countries – a mix of “modernization, marketization, minimization, and maintenance� to use their characterization. Political systems can either respond strategically, or use a muddling through approach which typically results in an incrementalist approach to reform. The conditions for conceptualizing and implementing a well-conceived strategic approach (a single coherent political authority, an informed leadership, ability to sustain change over a 5-year or longer period, capacity to organize and manage reform, and retain public support) are very rarely possible and that most actual reforms conform more to the “muddling through� approach based on partisan bargains and adjustments. This is all the more likely in a developing country context, where coherent political leadership and coordination on institutional reforms is a harder find. They then suggest that the endurance of strategic models of reform, a staple of the literature, is more of a conceptual device than a practical plan. 2 “From an incrementalist perspective, therefore, the nature of public management reform is “bitty�, ad hoc and specific, not strategic, comprehensive and driven by generic models. Models may sometimes still play a headline role, but from this perspective, they are still being used as a “selling angle� for something much more modest, or as a post hoc rationalization for the same. Their use, in short, is more of a rhetorical device than as a practical template for action.� (Pollitt et al. 2000) 2 If in fact most reforms are incremental, then the use of broad models of public management reform, whether focused on the budget system or the civil service system, are unlikely to be successful. In a review of public sector governance reforms from 2001 to 2011, Z. Scott points out the relative low success of public sector reforms and call for a “complete rethinking� (Oxford Policy Management 2011). Andrews (2013) identifies the problem as due to a tendency to design “square peg reforms for round hole governments� and attributes this to a desire to apply reforms that signal to an external audience that the government is worthy of external financial support. Such reforms pay inadequate attention to local context and typically fail to improve functional performance of government. The idea that the design of an institutional reform must be customized to the context is almost self- evident, yet the practice seems to suggest that this fundamental requirement is usually ignored. “Specialists� in public management reform use standard approaches to what appear to be standard technical public management challenges because understanding the context is a tedious and difficult affair, involving layers of complexity and requiring a level of judgment that is not often available. Yet project designs are complex in developing countries, well beyond the capacity of country authorities, and we might speculate that is often the reason why implementation falters or fails. The concern is that a systematic process of over-selling comprehensive models of public sector reform to developing countries is bad at many levels – it results in bad institutional design advice which, down the line, contributes to impact evaluations that are judged against the overpromised impacts and therefore results in systematic underperformance relative to an unattainable standard. And it is bad because it distracts from efforts at more modest incremental reforms which might be the only attainable path to improved governance and public management. What might an alternative approach to reform look like? A growing literature is centered around “pockets of effectiveness� in public sector reforms. Pockets of Effectiveness are public organizations, which provide public goods and services relatively effectively in a hostile environment dominated by poor management and bad governance. This approach strives to explain why pockets of effectiveness happen, what are the conditions for them to prosper (Roll, 2011 and Crook, 2011). Andrews (2013) suggests that “problem driven iterative adaptation� is more likely to lead to real functional improvements in government performance. Problem identification, engagement with multiple 2 Pollitt and Bouckaert (2011) update this perspective by noting that, “big models are stimulating, and will surely keep coming …. . Yet they do not take one very far, either in analyzing actual reforms or in prescribing what might be done in future. Worse, they can greviously mislead their users into thinking that they really know what is happening in country X or Y … (p.220). 3 stakeholders, step by step experimentation, and political acceptance are more likely to lead to institutional solutions to problems that will be adopted. Another approach can be traced to the “Innovation for Successful Societies� (ISS) program at Princeton University which collects case studies of institutional change, relying on first person interviews of participants in institutional reform. The ISS program provides the following logic for their approach: The two main strands of scholarship on institutions have both focused on sources of stability, not change. Rational choice approaches usually model institutions as equilibria, while historical institutionalist accounts often point to forms of path dependence that lock countries into particular organizational forms, policies or ways of doing things. In some of our work, we use case studies to help generate propositions or theories about why some reformers in some settings are able to depart from existing structures, rules, or norms while others are not. That is, we use some of our case studies as the basis for inductive generalization 3. The approach taken in this paper follows the ISS logic in that it looks at particular cases of institutional change in Cameroon to derive some general insights about how change happens. The approach also has some similarities to the experimental method described by Andrews and adds an explicit discussion of the political and bureaucratic context. But we also add an element of what we consider to be a useful innovation. In describing why one approach worked and the other did not, we turn to a discipline which has hitherto not been widely used to describe the processes that drive or block change in institutions: behavioral economics. The burgeoning field of behavioral economics focuses to a significant extent on how individuals respond to information and incentives, and challenges the notion that individuals are always rational and always maximizing utility. The suggestion is that in fact individuals often lack information or are not always consistent in their behaviors and often act in ways that deliver a perverse result relative to the expectations of conventional economics. Specific ideas, such as prospect theory, loss aversion, and the fallacy of planning have been developed by Kahnemann, which potentially have rich application to explaining behavior within large bureaucracies (including development agency bureaucracies) and in political systems. Despite a large literature on public management, little attention has been given to the problem of behavior within public sector organizations and its implications for the problem of public sector performance 4. Other than some conventional thinking on pay incentives and performance and some recent work on the politics of public sector reform, there is little in the literature to explain why so many public sector reform attempts fail. Lack of political commitment is a typical explanation but that begs the question, why initiate a reform? More likely the answer requires us to unlock the nature of behavior 3 See http://www.princeton.edu/successfulsocieties/research/more-on-case-studies/. 4 There is the classic work of Oliver Williamson, and Jensen and Meckling on organizations and agency and control in profit seeking firms but less work on government organizations. A search of the Google and SSRN database turns up only a reference to Behavioral Economics perspective on pension plans. 4 and collective action problems within government organizations and agencies 5. Addressing this problem requires an acknowledgement of the difficulty of institutional reform and the inertia, vested interests in the status quo, and collective action failures that often block reform and change. Starting with the hypothesis that behaviors are the critical (and often overlooked) factor in public sector performance, this paper explores the notion of how behavioral change (and thus institutional change) might be better motivated in the public sector. The basis for this study is “an accidental experiment� resulting from the World Bank’s operational engagement in Cameroon which provides some insights on how institutional change happens. In 2008, after almost four years of stop and go dialogue with the country authorities, World Bank staff successfully concluded preparation on a project to support the government of Cameroon to improve transparency, efficiency and accountability of public finance management. The US$15 million “Transparency and Accountability Capacity Building Project� project supported a number of Ministries to strengthen systems and capacities by upgrading the legislation and the budget, human resource, and procurement management and oversight systems. It was a conventional “investment lending project� and expected to achieve its objective in five years, a typical lifespan for a planned Bank project. Independently and concurrently, other Bank staff initiated a low-profile, technical assistance project to improve performance in Cameroon customs supported by a small trade facilitation grant of approximately US$300,000. The two projects of different scale, scope and design in the same governance environment offer a very interesting natural experiment (unplanned but accidental for that reason) that allows insights into the nature of institutional change, and the role of behavior and incentives and approaches that offer greater prospects for making reform possible. One approach appears to have succeeded in initiating change while the other has signally failed. What can we learn about institutional change from this accidental experiment with contrasting approaches and different outcomes within the same broad governance context? 2. The governance context in Cameroon Cameroon has experienced poor governance for much of its independent history. Repressive governance under its first president, Ahmadou Ahidjou, was followed by the rule of Paul Biya who has served as president since 1982, winning re-election to an additional seven year term in 2011 after amending the term limit provision of the constitution in 2008. The Mo Ibrahim Governance Index for Cameroon ranks it 36th out of 52 countries in Africa, with a stagnant score over the past six years. Cameroon scores particularly poorly on sub-scores for Rule of Law and on Participation and Human 5 See blog on problems in a typical developing country: http://blogs.worldbank.org/africacan/public-sector-reform- changing-behavior-with-cars-and-computers. 5 Rights. Cameroon topped Transparency International's list of the most corrupt countries in the world in 1998, 1999 and 2002. Reflecting this problem, Orock et al. (2012) noted, “corruption and embezzlement of state funds by public officials have become so endemic that they have come to be seen by most Cameroonians as the major knot that ties together their problems with poverty, unemployment and insecurity�. In terms of the functioning of the bureaucracy, a 2009 Development Policy Review by the World Bank 6 noted that “uneven commitment to reform, inconsistency and pervasive bureaucratic inertia distinguish Cameroon from many high performing developing countries.� It pointed out that: “bureaucratic inertia …has undermined its capacity to effectively execute policies.� It pointed to “the root cause of this pervasive problem, namely the excessive fragmentation of the administration and the proliferation of overlapping agencies (ministries, permanent committees, special commissions, and so forth); the patronage based promotion and demotion system, which contributes to low morale and ineffectiveness in the civil service; and the indiscriminate low pay policy which seriously undermines the emergence of a stable and well-motivated technocracy in Cameroon administration.� Others have commented on the effects of a patronage-based system of appointments and demotions on the performance of the bureaucracy. 7 The net effect is a high degree of reform inertia, low levels of mutual trust, and weak coordination across ministerial boundaries, precisely the symptoms identified by the report. 3. The first story: The transparency and accountability project Project design and results In June 2008, a conventionally designed transparency and accountability capacity development project (TACD) was approved by the World Bank board. It consisted of a USD 15 million International Development Association (IDA) credit in support of a fairly ambitious and comprehensive program of reform across three ministries. The project development objectives (PDO) were: (i) to enhance transparency and efficiency in the public finance management (PFM) of Cameroon; and (ii) to strengthen accountability in the use of public resources in Cameroon. The PDO was to be achieved through four project components: the first component of the project was to improve budget management including procurement. In this component, focus was mostly on giving support to revision of processes, methods and regulations for budget programming, budget execution including procurement, budget monitoring and controls, and treasury accounting in the context of the new budget Organic Law. The second component of the project was about integration and development of the Financial Management Information System (FMIS) including the Human Resources Management (HRM) and Payroll Management Systems. The third component of the project was to strengthen 6 “Sustaining Reforms for Inclusive Growth in Cameroon: A Development Policy Review�, World Bank, 2009. 7 See Henry Kam-Kah, “The Culture of Appointments, Arrogance and Chop Broke Pottism in Cameroon’s Contemporary Governance�, http://www.cjdhr.org/2010-12/Henry-Kah.pdf 6 external oversight and monitoring and evaluation (M&E). The fourth and the final component of the project was support to the project focal point for PFM reforms coordination and capacity building activities for Ministry of Finance to properly carry out their coordination, monitoring, and leadership roles in PFM reforms. The project was anticipated to be completed in five years, with expected disbursements of $ 1.8 million, 3.2 million, 4 million, 3.5 million, and 2.5 million. This project only became effective one year later, in 2009, itself a warning sign of problematic commitment to the project objectives and weak ownership of the project. After two years of implementation, only 10 percent of the total budget had been disbursed relative to the 33 percent projected at the start. The project received high level management attention when a visit by the World Bank Vice President of the Africa region to Cameroon included explicit discussion with the political leadership of the country and the ministry regarding the need for effective project implementation. When this failed to galvanize action, World Bank management conveyed the Bank's concerns with the status quo regarding the project. The main institutional counterpart in government, the Ministry of Finance, indicated their intention to strengthen implementation and closer supervision by the Bank ensued. A Road map was then adopted and by the Mid-Term Review carried out in August 2011 it was found that the project had been partially rectified. However, the project lagged in many respects and after further consultation with the government a mutually agreed decision was made to close the TACD project in December 2012 (one year before its scheduled close) and to reallocate the undisbursed amount to another project. The documentation on this project is very revealing about the dialogue with the government and the various stages of decision making that typically accompanies such a project in a typical Bank-client relationship. The project preparation itself was unusually drawn out with the concept preparation starting in 2004 and finally being approved in 2008. World Bank supervision missions worked with government counterparts to try and achieve the development objectives of the project but reported slow progress in most components of the project. The reasons were myriad but a typical set included: (i) delays in building up the PFM reform consensus within the Ministry of Finance and the government, (ii) organizational deficiencies in the Ministry; and (iii) lack of leadership and organizational skills. This mainly affected the project procurement processes leading to delay in starting critical activities in the 2009 project work plan, such as the IT upgrading and the payroll system audit and upgrading. 8 In addition, the supervision team noted that: 8 World Bank Supervision mission report, November 2009. 7 “Five months after effectiveness, the Project Unit is still lacking some basic facilities and organizational empowerment that were officially pledged during past visits. The impact of the PFM modernization program and associated projects very much depends on the willingness of the Ministry of Finance to lead the process and own the program, which commitment remains in doubt. Absent prompt actions to accelerate implementation of project activities, the Bank team regards the risks to the project development and sustainability as very high.� A subsequent supervision in May 2010 noted that there had been some progress on budget-related components but made the following observations about weak coordination across the involved ministries: Poor coordination between the Ministry of Finance and the Ministry of Civil Service explains the delays in launching the HR and Payroll audit which is a critical preliminary step towards reining in the wage bill, and improving the public sector management. Weak coordination between the Ministry of Finance and the Ministry of Plan, and within each department, is hindering the initial efforts deployed in reforming budget formulation and preparation. In terms of formal project supervision rating, TACD was rated Marginally Unsatisfactory (MU) for progress towards achievement of the development objectives for the first two years and was then downgraded to the lowest possible rating of Unsatisfactory (U). Despite high level dialogue between the relevant Ministers and World Bank management similar problems continued to bedevil the project over the next two years. The general litany of problems that affected this project would typically be put under the catch-all category of “a lack of political commitment� by the government to undertaking the necessary actions. But the real problem is often more subtle and nuanced and deserves more analysis. This paper only goes so far as to offer some hypotheses about the reasons why such projects fail. The reasons have much to do with the design of public management reforms and their failure to adapt to the context. Political context matters very much but its influence is something that should be factored into the design. Given a specific authorizing political context in the country, greater effort needed to be expended on identifying the kind of “incremental change� that is most likely, given the likely resistance to reform and the relative strength and capability of champions of reform within government. With hindsight, there was little investment in undertaking such an assessment of the authorizing context and therefore the project may have had fundamental design features that were very unlikely to be achievable in Cameroon. For example, the project required a fairly significant amount of coordination across the three ministries involved in implementing the components of the project. However, such coordination failures were predictable in the political and bureaucratic context of Cameroon but were not specifically identified during project preparation and design. With regard to design of public sector reform, broad models of transparency and accountability tend to block the necessary thinking with regard to likely institutional change in a particular country or 8 institution at a particular time and the result is fairly standard approaches to systems reforms which ignore the context of “time and place�. But most importantly, they ignore the fact that institutions are about people, about incentives, and about behaviors. Why is behavioral economics relevant? How does behavioral economics offer a way to think about incremental change or the limitations to broad systemic change? The first hypothesis is that the principles of behavioral economics apply to both parties to this interaction – the development agency offering technical advice and assistance (in this case the World Bank) as well as the various government counterpart agencies. A number of behavioral economics concepts appear promising in explaining behaviors that at first glance appear to be irrational. First is the concept of planning fallacy – which is the tendency of people and organizations to underestimate the time it will take to complete a task or achieve an objective, even when past experience points to more time required. Much of public sector reform design exhibits this planning fallacy despite a mountain of evidence that indicates that projects take longer to implement than projected. Development agencies are particularly prone to this problem and the behavioral reasons for this are worth investigating. Behavioral economics offers one possible explanation:� the "authorization imperative" which is described as follows: “Much of project planning takes place in a context where financial approval is needed to proceed with the project and the planner often has a stake in getting the project approved. This dynamic may lead to a tendency on the part of the planner to deliberately underestimate the project effort required. It is easier to get forgiveness (for overruns) than permission (to commence the project if a realistic effort estimate were provided.) Such deliberate underestimation has been named strategic misrepresentation.� 9 A close cousin of the planning fallacy is “optimism bias�, which tends to the belief that despite the evidence of past projects, the current project is somehow exempt from the risks that affected previous attempts. This too is a strong behavioral pattern that may be at play both in development agencies and in government agencies. A third important behavioral economics concept is the idea of “loss aversion� – which refers to individuals strongly preferring to avoid a loss over the prospect of making a gain. This concept has potentially powerful application to explaining why individuals and institutions appear to be so resistant to change. Most reforms involve change and change is perceived to create risks both to politicians and 9 http://en.wikipedia.org/wiki/Planning_fallacy. 9 to bureaucrats. Managing such risks and avoiding a loss, of power, of prestige, and command over resources is then a major motivation of those directly impacted by reform and those steering reform. When the consensus regarding reform is limited, these reforms appear to be highly risky with any misstep likely to be punished by dismissal and replacement by a rival. In an authorizing political environment such as obtains in Cameroon, major reforms are likely to be highly risky for senior technocrats and ministers. Mutual trust between key players is also undermined and enhances the sense of risk and is reflected in poor coordination and collaboration across key ministries involved in reform. All of these factors are observed in the experience of the TACD project and may explain why a project that required high level political commitment and coordination failed to achieve effective implementation. This explanation offers some useful insights on why broadly conceived public sector reforms may be poorly designed for a context where the necessary behaviors are unlikely to be generated to make the reform successful 10. If loss aversion is a general behavioral principle that affects individuals and institutions, then we will have to factor that into project design and see how the risks of reform can be addressed. The approach taken to the customs reform in Cameroon provides an illuminating example of how behaviors could be factored into the design of reform even in a difficult authorizing environment. 4. The second story: Pilot reforms of Cameroon Customs Customs reform context in Cameroon The function assigned to customs services is to collect duties and taxes on behalf of the treasury. In Cameroon, the share of taxes and duties collected by customs amounts to 27 per cent of the national budget (Cantens 2012). However, there is evidence to suggest that the taxation of dutiable goods is higher than official figures and that some proportion of the revenue is diverted to private purposes. Additionally, there is evasion of customs duties with the active collusion of customs officials (Cantens et al. 2012) 11. “African customs officials are perceived as a national elite involved in development and dragged into corruption.� (Cantens 2012). Cameroon customs was considered as one of the three most corrupt national administrations. Customs officers themselves admit that corruption is a major obstacle to any reform (Libom et al. 2009). 10 Rent-seeking is also another explanation of the TACD problems. Some Ministers and senior level technocrats competed around rents and profits extracted/expected from the IT contracts. 11 For a description of corruption extent and practices, see Cantens et al. (2012). 10 The pay structure in the customs department involves a high structural differential between the lowest and highest salary grades, as well as a differential due to the incentive system which provided bonuses on the basis of duties collected. In Europe, the income differential between highest and lowest ranks of customs officials does not exceed 1 to 10. In Cameroon customs, this difference is roughly at 1 to 50, and perhaps 1 to 100 in the case of efficient customs investigators 12, and even greater for corrupt officers (Cantens 2012). In principle therefore the financial incentives for good performance were strong. However, salary and incentives did not seem to ensure good revenue performance or to restrain corruption in customs. Neither were skills and capabilities a reason for poor customs performance. Indeed, Cameroon customs officers are very well trained at the ENAM (the best ones usually select customs when they exit the competitive civil service curriculum). In addition, Cameroon customs collaborates with five major international organizations and donors on modernization initiatives. This means there were at least one or two meetings per month between senior officers and international experts. Thus, customs officers have no difficulty imagining what services a modern customs service must perform (Cantens 2012). Attempts to improve performance through previous reforms had not been successful. A major problem was that most reforms usually did not cascade to the level of frontline officers because there is an information asymmetry between the head of customs and frontline officers. The DG of Customs felt that it was necessary to strengthen the hierarchical link and provide the head of customs with more information on practices on the ground13. At the request of the Director General of Customs, the World Bank offered to support some pilot reforms that would begin to experiment with better performance management of the department. The experiment was designed based on detailed knowledge of practices since the World Bank had been involved in customs reforms since 2005-2006. In addition, the World Bank worked closely with a counterpart from the World Customs Organization, a former French customs officer, who had been adviser to the head of Cameroon customs from 2006 to 2009 and brought detailed institutional knowledge and relationships. The diagnostic work undertaken before the Bank intervention and the knowledge of the customs issues gathered over these 4 year period by the French technical assistance project was crucial in framing a more adapted and sound reform. The performance contract experiment built on the IT modernization initiative. In 2005, the Cameroon government undertook the implementation of a new IT system at the port of Douala by 1 January 2007, following an IMF recommendation. Cameroon customs senior officers did not share a common view on outcomes and impacts on the customs service. For some of them, this new IT system, which was provided by an international organization, had to be adapted in order to fit exactly with the existing procedures and the procedures themselves to be changed only slightly. Others, the reformers, 12 Customs inspectors are allowed to keep a couple of percentage of any fraud. This is a legal incentive for them to discover and sanction fraud cases. 13 This is explained by the head of customs in Libom et al. (2009). The same phenomenon was also observed in a low governance environment, DRC, with the case study of Kasumbalesa border-post (Cuvelier et al. 2013). 11 supported the idea that a new IT system provided by an international organization was an opportunity to reform the procedures drastically, and for the Cameroon customs to achieve international standards. Ultimately, some ‘reformers’ were appointed to manage the project and they succeeded in reforming some customs procedures after numerous clashes (Cantens 2007). Performance contract design and results A pilot program of performance contracts was then implemented in Cameroonian customs and has been under implementation for over three years. It started with performance contracts for a six-month period in 2010 for two customs offices and was then expanded gradually to two other offices and then importers 14. The pilot program was the first of its kind and no similar project has been carried out in fiscal administrations 15. It was limited to knowledge transfer and technical assistance with no provision of equipment/computers and formal training. The project was funded by a trade facilitation grant of approximately US$ 300,000. This was designed accordingly for several reasons: the first reason is that after several years of support to Cameroon customs reforms (it had started at the end of the 90s within budget support operations and resumed with project lending in 2007), there was a feeling that lending, either through budget support or project lending, had brought some perverse incentives on some customs officers and had been largely counter-productive. Therefore, relying exclusively on technical assistance was felt as a way to go. It is worth noting that the focus was on non-financial incentives since it was felt that the extent of corruption was so high for some officers that financial incentives could not offset illegal payments. Moreover, financial incentives are usually less effective at changing actors’ behaviors than financial disincentives. The performance contracts program formalized an agreement between the customs managers and the customs officers, and had at its core a system of non-financial incentives and sanctions to be applied to officers’ performance. The contracts were agreed with the individual customs officials, and set out in detail specific results for eight indicators that had to be met. Importantly, they went beyond the revenue targets fixed by the government to include organizational and best practices issues. The TA project in Cameroon customs also included very thorough work to develop internal tools to inform management of individual performance, helping tackle issues of information asymmetry. 14 For detailed description of the results and an historical perspective, see Bilangna et al. (2012). 15 Customs is an area of public sector reform with several positive factors mentioned in Crook (2010): (i) sense of commitment to a mission on the part of employees, associated with the belief that they were doing a valued and useful job – hence pride in their job, (ii) a sense of being part of a team, an enjoyment of the solidarity among work mates, (iii) managers who demanded, recognized and rewarded good work performance, (iv) A sense among employees that they were special and had been selected for their competence, (v) relative organizational autonomy in operational matters. 12 The program defined that customs agents that were successful in meeting their performance goals would receive mainly non-financial incentives including congratulatory letters that were put in the official’s permanent file and disseminated publicly to provide wider recognition as one of the best inspectors in the customs service. Other incentives included options to participate in training courses, and other similar measures. With regards to sanctions for bad behavior, the program introduced a process of interviews with the program team when the performance contracts commitments were being broken, and warnings as to possible disciplinary action that could be taken if performance did not improve. The main sanction was removal from their position, as in the targeted offices, the possibility of earning money legally through disputed claims was high (since Cameroonian customs agents receive a share of any revenue obtained from fines levied on claims). The concrete sanction of being transferred to an office with lower earning potential was a financial disincentive that was considered to have a greater impact on personal behavior than a positive financial incentive. This illustrates the use of the principle of loss aversion as a strong motivation for changing behaviors. As result, an increase in the volume of declarations processed generated an additional $16.5 million in revenue over the six month period. The effect of the contracts program on trade facilitation was positive. Prior to the implementation of the program, less than 80% of declarations were assessed on the day they were registered, compared with almost 90% during the program. Broadly speaking, three non-quantifiable results have been observed. First, performance contracts made it easier to conduct inspection procedures with the operator of the container terminal. Second, customs agents became more “diligent.� The performance contracts induced agents to stay longer in the office—time at work increased by between 3% and 10% in the two offices, whereas other Cameroonian offices that were not part of the program had time at work decrease by -4%. Third, customs officials became more aware of their responsibilities, improving the relationship between customs agents and their managers. The improvements in customs processes have also been noted by private sector actors. At the start of the program only 17% of respondents from a survey of brokers had a positive perception of the customs agents participating in the program; these same companies had a very positive perception of the customs agents at the conclusion of the program (other 90%). According to the survey, the most important impact of the program has been the speeding of the clearance process and the better attitude of customs inspectors. As a result of the program, in less than two months the behavior of participating customs inspectors changed significantly for the better, leading to reduced clearance time, increased revenues for the Cameroonian Treasury, and a reduction in harmful corruption-related practices. Moreover, the pilot was extended to two other offices and to large importers a year after. It is always difficult to attribute without a doubt, success to such experiment. However, what could be easily perceived (confirmed by interviews of private operators) was that the “reformers� in the organization were strengthened and some of them promoted. Based on positive results in terms of revenues collection, it became a little bit easier for the head of customs to get approval from the 13 Minister of Finance to promote the best inspectors (since it was based on objective criteria) and had more leeway to change/sanction inspectors since it was again based on objective and transparent criteria (which had never been really the case before). Utterly deviant practices reduced significantly since the likelihood to be sanctioned became higher and customs statements on reforms became increasingly credible for some compliant operators and act as a deterrent for the most non-compliant importers. Some interviewed informal importers explained that because of the reduction of profits due to customs changes, it had become more profitable to leave the trading business. This is therefore a virtuous circle, which has been increasingly put in place. It is however worth noting that some non- compliant practices obviously continue. One could argue that these positive changes could have been of short-term with a gaming effect. However, over the final years, from a revenue perspective, customs reforms in Cameroon seem to be sustained 16 (see Figure 1). Between 2009 and 2012, revenues have increased by almost 0.3% of Cameroon GDP. If revenues would have remained at 4.45% like it was in 2009 before the experiment, Cameroon customs would have lost in 2012 30 billion FCFA or approximately 60 million US$. It is worth noting however that when revenues have increased, exemptions (which are usually signed by the Minister of Finance) have even increased more rapidly: they were equal to 0.6% of GDP in 2007 and increased to 1.5% in 2012 accounting for almost 1/3 of collected revenues. Figure 1: Customs Revenues and Exemptions as share of GDP in Cameroon (2007-2012) 8.00% y = 0.0028x + 0.0435 R² = 0.9222 revenues+exemptions/ 7.00% GDP 6.21% 5.67% 6.00% 5.22% revenues/GDP 4.89% 5.28% 4.71% y = 0.0012x + 0.04 5.00% R² = 0.9317 4.54% 4.58% 4.68% Linear 4.00% 4.45% (revenues+exemptions/ 4.10% 4.17% GDP) 3.00% Linear (revenues/GDP) 2.00% 2007 2008 2009 2010 2011 2012 Data source: Cameroon customs and WDI. Finally, was the success of the Cameroon customs experience due to the personality and commitment and change management skills of the head of Cameroonian customs? The head of customs has had definitely a positive impact. However, this would be too simplistic to attribute all the success to the 16 Whereas customs clearance time remains stable with more than 90% of declarations processed the same day declarations are lodged. 14 head of customs since a parallel activity with Cameroon customs was not successful either. Indeed, during the same period of time, within the framework of a World Bank supported lending project 17, which included procurement of equipment and formal training, several activities, such as connecting two main border-posts to the HQ IT system has been lagging for several years (whereas the main counterpart on the Cameroonian side is the same person: the head of customs). It is worth noting that a lending project to Customs has not led to any major change on the ground but a new technical assistance activity, using mirror statistics (within the framework of performance contracts) has led to revenues increase, notably for imports of motorcycles. What has been probably critical has been years of dialogue and getting to know from inside the administration as well as the gradual buy-in of a key number of agents below the head of customs. 5. Inferences from the Accidental Experiment with two different approaches At the risk of generalizing from these two case studies, the following inferences could be drawn from this example. First, one could hypothesize that the small size of the customs project and a lack of financing of equipment and of studies was a good fit with the counterpart’s ability to authorize the necessary actions. The DG of Customs was able to provide the leadership of the project without the need for layers of approval from the political authorities, something which is very difficult to obtain in the governance environment in Cameroon. By contrast, the conventional project required a level of coordination and cooperation across ministries that could not be provided even by the Minister of Finance in the political environment of Cameroon. Rivalry among ministers ensured that collaboration and coordination would be problematic and was not adequately considered in the design of the project. This problem bedevils most conventional project designs. A project implementation unit was created and staffed up, but its performance was uneven, reflecting the need for the head of the PIU to negotiate actions with multiple ministries, including his own Minister (MoF), whose leadership was typically uncooperative. Second, the customs project did not seem to suffer from the fallacy of planning. Focusing on the changes in behavior at the pilot level brought a level of attention to that problem which is normally not addressed by conventional project designs. The expectation in conventional projects is that a project implementation unit will resolve planning problems and implementation matters when in fact this is a difficult change to achieve in most bureaucracies, the more so in weak governance environments where inertial forces are very strong. Attention to change management, careful planning and the subtle and non-subtle changes to incentives and behaviors in institutions is often neglected or under-served in conventional projects. Even where there is a focus on change management, it is difficult to conceive of 17 The CEMAC trade and transport facilitation project. 15 changes in behavior across the full scope of a conventional project which is typically very broad. So most reform “components� languish and are not implemented. Third, the customs project, by the nature of its size and flexibility, also was not burdened by the expectations and compliance overload that is attached to a conventional project. For the head of Customs, if the pilot had failed to improve performance, the loss would have remained limited. The scale of the project therefore also enabled better management of the risks from reform, particularly one which involved a novel approach to performance management. A pilot approach is therefore useful to tackle a loss aversion problem. Fourth, the approach to customs reforms worked in part because it was a genuine partnership between the head of the agency and her key staff and the World Bank and World Customs Organization staff. The counterparts on both sides stayed the same, avoiding the changing of faces that accompanies many projects. A considerable amount of mutual trust was built up on both sides based on cordial but candid communication. The external partners did not approach the government with complex reform ideas but provided specific technical advice to enable identifiable problems to be solved. Where the solutions were not obvious, an experimental approach was tried with careful monitoring of the results and recalibration as needed. Without consciously meaning to follow good practice advice, the scope of the reform was kept simple and achievable and incremental gains were sought. 18 The experience suggests that with institutional reforms, implementing a series of small well-designed changes may hold more hope for behavioral change than a large but ineffective reform that presumes the capacity for internal leadership of a complex reform. Where the reform design is not self-evident, experimental approaches may hold more promise than is commonly acknowledged. Equally importantly, a low-key approach enabled the Cameroonian counterparts to master the techniques, to reach achievable goals, and become more self-confident in taking ownership of the reforms. As small successes became evident, the government team was given due credit and given opportunities to present their approach at international fora, providing strong reinforcement for their reform efforts. As this happened, the concerns about loss aversion was gradually replaced with greater self-confidence and a change in behavior and aspirations for performance. The circle of reformers in customs gradually expanded to include more and more staff. This critical achievement bodes well for the long term sustainability for reform. The scale and flexibility of the project also benefited the World Bank staff member managing the project. The WB-WCO team was able to focus on the issues and work closely with counterparts to design experiments that targeted behavior change and not worry about reporting back to a bureaucracy on disbursements and fiduciary safeguards. Natsios (2010) provides a compelling discussion of the problems created for frontline operations because of the compliance demands imposed by the “counter-bureaucracy� which dominates the decisions of development agencies such as USAID and the 18 This is in line with the literature which recommends a focus on incremental changes (Politt et al. 2000, Oxford Policy Management 2011). 16 World Bank. The small grant operation was able to function without the full burden of the counter- bureaucracy and as a result, the task team was able to be innovative, focus on the constraints on behavior and work below the “compliance� radar to achieve operational impact. This highlights the need to overcome some of the behavioral problems within the development agency itself which often leads to project designs that even task team leaders (TTLs) have little control over and which makes implementation almost impossible. There is a trade-off between a heavy focus on procurement and fiduciary compliance (the probability of committing error type 1, of having a corrupt transaction in a high profile project) and the ability to design a project that is responsive to the problem at hand (error type 2, having a low profile project). Over-emphasizing the former will result in staff who are skilled at compliance but who fail at designing and implementing projects that achieve impact. The reason development agencies do not undertake more small, focused projects is because the decision rule that guides agencies seems often to be focused on delivering the most aid dollars at the least administrative cost. This leads to a false sense of economy which results in pressures to create large high value projects with ambitious reform objectives 19. This is the problem of the “authorization imperative� so that the scope of the project design is systematically biased towards the large and ambitious project which is a characteristic of the conventional project. The environment in large development agencies thus is inimical to the small innovative project which experience indicates is more likely to have impact and change behaviors in reforming organizations under poor capacity/governance environments. The moral hazard effects of lending should be explicitly factored into the design and approach. In a context of poor governance, counterparts have a strong incentive to misrepresent their commitment to reforms when the project is seen to offer access to significant financial resources. Most public sector reforms are about management and organization and many problems could be addressed with non- financial advice and support. Lending should be tailored to explicit needs and overfunding of projects should be strenuously avoided since it usually creates negative distortions. 6. Potential issues of replicability and concluding thoughts Even though an experiment could probably be successful in any country (if well targeted and designed), it would not mean that it would have a “catalytic� effect and this could take several years to materialize. However, in traditional PSM projects, the same “catalytic� effect may not materialize either since the impact on the ground is sometimes problematic, which hampers the change agenda. As Roll (2011) pointed out, pockets of effectiveness “perform relatively well but are not a magic bullet for transforming the public sector either. Because they perform essential functions in ‘bad governance’ contexts, they are still important enough to be studied and supported on their own merits�. 19 Some of the activities of the TACD could have been launched when the closing decision was made to reduce supervision costs and to present a better picture of the lending portfolio. 17 Several factors related to the political economy of international organizations may limit the replicability of the Cameroon customs approach. Donors are often criticized for being “short-termist� in their approaches and favoring ‘quick wins’ in project selection and design rather than supporting long-term, systemic reform. The reason for this behavior is attributed to donor agencies being under pressure from their domestic constituencies to demonstrate results within relatively short budget cycles (Oxford Policy Management 2011). Moreover, solutions are idiosyncratic, which hampers the possibility to replicate any solution. Only the approach is replicable, not the solution. However, several experts argue that donors usually import a “standard, foreign model of reform� (Oxford Policy Management 2011). Such approach does not lead to the constitution of core indicators easy to monitor for having a “macro� view of public sector projects. Finally, in organizations which organize around lending operations, which have perverse effects, this approach may sometimes be resisted. Moreover, coercion usually does not foster change (Andrews 2011). Based on the impact evaluation from this natural experiment, this paper presents a case study that the smaller scale, flexible approach focused on a specific agency was more effective than a broadly defined conventional approach. The results, which will need to be replicated in other contexts to convince the skeptical, have very significant implications for how organizations such as the World Bank undertake public sector reform. The conventional project demonstrates an observation that is repeated across many countries; that it is difficult to successfully implement a public sector reform project with a design that calls for cross- government coordination of reforms where the authorizing/governance environment is not supportive. The lesson learned here is that understanding the political economy and the authorizing environment is critical to ensure that project design suits the political context. By contrast, the smaller Cameroon customs reform was a reform that was scaled to better fit the context; allowed the Director General of Customs to select a set of pilots to manage the risks of reform; and enabled a stronger process of learning and building a trusting relationship with the staff of the World Bank and World Customs organization. This reduced the problem of loss aversion. After 5 years, this process has now resulted in a very successful set of reforms on customs performance management with significant improvements in customs revenue collection and private sector satisfaction with the customs process. More importantly, the DG of customs and her staff are now a very self-confident set of reformers who have made presentations on the Cameroon customs reform approach to international fora, including at the World Customs Organization and at the World Bank. There are a number of lessons that could be drawn from these two experiences which is that project design is a critical element of a successful reform. Too often the Bank's conventional project preparation approach involves a fairly large scale of reform with a number of components that quickly adds up to US$20-30 million. Typically, this is well beyond the management capacity of the government 18 counterparts and the TTL also is weighed down with complex project management requirements that take focus away from problem solving. The project design is cast in stone and can only be modified through a complicated and bureaucratic process of restructuring the project. The customs project experience suggests that both the scale and the design may need to be more modest in order to actually spur reform. But perhaps the most important insight to take away is that public sector reform is about behavioral change and that requires internal leadership which needs to be carefully identified and supported and encouraged to proceed at a pace which is self-defined. External support must then be flexible and paced to the process of change that is internally acceptable. The use of behavioral economics provides useful insights into understanding both the behavior of development agencies in persisting with an approach to public sector reform that has generated a low success rate, as well as explaining the strong inertial tendencies in most public sector systems to change. With regard to the former, the issue of the authorization imperative creates strong incentives for behaviors which reinforce problems of the fallacy of planning and the optimism bias. Where the focus is on ensuring that resources are committed to projects, the authorization imperative creates a repeated game with very little learning and improvements to avoid the mistakes of the past. With regard to the inertia in public sector institutions, a key idea that emerges is the relevance of the idea of loss aversion as a strong factor in explaining why governments that commit to reform often are unable to follow through in implementation. If loss aversion is based on perceptions of risk, an important operational principle that emerges is to assist governments to manage these risks, potentially by undertaking pilot reforms which allow the implementers the ability to control the risks while developing the capacity to handle change. Given the acknowledgement that most public sector reforms are incrementalist, a necessary implication is that public sector reform may have to take the form of small-scale incremental pilots which over time strengthen the capacity of the actors to gradually reform the system. 19 References Andrews, M. (2013), “The Limits of Institutional Reform in Development: Changing Rules for Realistic Solutions�, Cambridge University Press. Andrews, M. (2011), “Does Change Space Influence Public Sector Management Project Success�, Mimeo. Bilangna, Samson and Djeuwo, Marcellin (2012), “The Figures Culture in Cameroon Customs� in Reform by Numbers, Cantens, Thomas, Ireland, Robert and Raballand, Gaël (eds.), Washington: World Bank Cantens, Thomas (2007), « La réforme de la douane camerounaise à l’aide d’un logiciel des Nations Unies ou l’appropriation d’un outil de finances publiques », .Afrique Contemporaine, 223-224/3-4, p. 289-307. Cantens, Thomas, Raballand, Gaël, Bilangna, Samson and Djeuwo, Marcellin (2012), "Comment la contractualisation dans les administrations fiscales peut-elle limiter la corruption et la fraude ? Le cas des douanes camerounaises ». Revue d’économie du développement, vol. 26(3), pages 35-66. Cantens, Thomas (2012), “Is It Possible to Reform a Customs Administration? The Role of the Customs Elite on the Reform Process in Cameroon�. In The Role of Elites in Economic Development, ed. By Alice Amsden, Alisa DiCaprio and James Robinson. Oxford, Helsinki : Oxford University Press, pp. 281-306. Crook, R. C.(2010), “Rethinking Civil Service Reform in Africa: 'Islands of Effectiveness' and Organisational Commitment�, Commonwealth & Comparative Politics, 48: 4, 479-504. Cuvelier, J. and Muamba Mumbunda, P. (2013), “Réforme Douanière Néolibérale, Fragilité Etatique et Pluralisme Normatif�, Politique africaine n° 129. Kahnemann, Daniel (2011), Thinking, Fast and Slow, New York: Farrar, Straus and Giroux. Kam-Kah, Henry, “The Culture of Appointments, Arrogance and Chop Broke Pottism in Cameroon’s Contemporary Governance�, http://www.cjdhr.org/2010-12/Henry-Kah.pdf Libom, Minette., Cantens, Thomas and Bilangna, Samson (2009). “Gazing into the Mirror: Operational Internal Audit in Cameroon Customs�, SSATP Discussion Paper 8. Washington, DC: World Bank. Natsios, A. (2010), The Clash of the Counter-bureaucracy and Development, Center For Global Development Essay, available at: www.cgdev.org/content/publications/detail/1424271. 20 Oxford Policy Management (2011), Evaluation of Public Sector Governance Reforms 2001-2011, available at: http://www.gsdrc.org/go/display&type=Document&id=4229&utm_medium=rss&utm_campaign=gsdrc &utm_source=newsfeed. Pollitt, Christopher and Bouckaert, Geert (2000), Public Management Reform: A Comparative Analysis - Oxford University Press. Pollitt, Christopher and Bouckaert, Geert (2011), Public Management Reform: A Comparative Analysis - New Public Management, Governance, and the Neo-Weberian State, Oxford University Press. Rogers Tabe Egbe Orock & Oben Timothy Mbuagbo (2012): ‘Why government should not collect taxes’: grand corruption in government and citizens' views on taxation in Cameroon, Review of African Political Economy, 39, 479-499. Roll, M. (2011), “ Can ‘Pockets of Effectiveness’ trigger public sector transformation in Africa?�, Paper prepared for the 4th European Conference on African Studies “African Engagement: On whose terms?� in Uppsala, Sweden, 15-18 June 2011, available at: http://www.nai.uu.se/ecas-4/panels/61-80/panel- 69/Michael-Roll-Full-paper.pdf. Schick, Allen (1996), The spirit of reform: managing the New Zealand state sector in a time of change, Wellington, State Services Commission. World Bank (2006), Enterprise survey, Washington: The World Bank. World Bank (2009), Sustaining Reforms for Inclusive Growth in Cameroon: A Development Policy Review. 21