Abstract
The new World Bank strategy on Public Sector Management (PSM) has to be judged against the magnitude of the problems it faces. Two of these are what to do about failed and failing states and what to do about the stagnation of progress in ‘flailing’ states. On both of these fronts the gap between what the world of action wants – an implementable action plan to tackle tough problems like Afghanistan – and what the PSM strategy does (and quite probably can) deliver is wide. This article proposes ‘problem-driven iterative analysis’ as one way of bridging the gap.
Keywords
Introduction: the crisis of public sector management
The World Bank’s new Approach to Public Sector Management 2011–2020: Better Results from Public Sector Institutions is a remarkable, and revealing, document. It manages, in spite of the constraints of an official document of a large multilateral institution, to convey a sense of the deep frustration and nearly complete puzzlement as to what exactly the World Bank should do next. Conveying substantive content within the constraints of World Bank internal processes is like having to give instructions to assemble a bicycle as haiku. Criticizing this, or any World Bank document, for not being specific and concrete is like criticizing a Zen koan for not being clear.
Within the constraints of the form, this document reveals the deep crisis in public sector management today, as there is less and less faith that, for those countries for which the standard recipes have not yet worked (and the standard recipes have in fact worked, more or less to perfection, in some countries) that anyone knows, with any certitude, what will in fact work to create high capability and high performance public sector institutions. Yet solving the problems of public sector management is essential, and hence just avoiding the problem is also not an option.
As words are short and issues complex, in this article I will resort to several metaphors. The first is of a series of doors. To get through them you have been given a set of keys. On some doors, the first key you try works, and the door swings open easily. On other doors, the first key also fits. It has to be jimmied and worked and some oil applied, but it eventually opens the door. On still other doors, the first key doesn’t work at all and other keys have to be tried, but eventually the ring has a key that fits the door. However, there remain some doors where all the keys on the ring have been tried, all of them have been jimmied and oil has been applied, yet the doors remain stubbornly locked. Now what do you do? You could give up. You could get some dynamite and just blow the door out of the way. You could start to look for a window to climb through.
Areas of incapacity for PSM
Let me briefly discuss two situations that standard PSM is hard pressed to deal with, before coming back to some suggestions about where open windows (or at least windows that can be jimmied) might be found.
Fragile states
Afghanistan: the report never uses the word. Yet it is a perfect illustration of a deep problem. The simplistic story is that the wealthiest and most powerful country in the history of the world invaded a small out-of-the-way country. On the ‘if you break it, you’ve bought it’ principle, the country then decided that it should (re)build a capable state before it left. So the question of how to build a capable state became a first-order question for the most powerful country in the world. Now, more than a decade later, the US has spent more than half a trillion dollars in Afghanistan (about five years’ worth of total annual global aid spending). Instead of success preceding withdrawal, the goal of a ‘capable state’ has been watered down little by little. President Barack Obama’s speech of May 2012, announcing US plans for withdrawal from Afghanistan, contained no reference to any state capability other than security (and the Afghan security forces at the time were providing adequate security to only half the population).
Success in high-profile projects creates an organizational and professional aura that lasts a long time, whether the project was worthwhile or not (the Manhattan Project’s success in the 1940s in building an atomic weapon, and NASA’s success in putting a man on the moon). But the converse is also true (the failure of US armed forces in Vietnam, and the mishandling of the New Orleans floods). I visited Afghanistan in 2006. In the overall strategy for governance in Afghanistan, the World Bank had been assigned the task of civil service reform. Discussing the World Bank’s engagement in civil service reform at the time I had the distinct impression this was not like rearranging the deckchairs on the Titanic. This was more like rearranging the little umbrellas in the drinks in the cup holders on the deckchairs of the Titanic.
How does the reputation of PSM in the World Bank emerge from Afghanistan? The narrative of the US military is: ‘We created the space for building a capable state, but the development community couldn’t deliver, so we had to do it on our own. We admit that we are cut out for what we in the military call “clear”, not “build” or “hold”. The failure of state building is the development community’s failure.’ Their story implies that PSM experts don’t know how to build a state. Perhaps the rejoinder might be: ‘We PSM experts said the right things, but we weren’t listened to.’ Perhaps the story is, alternatively, that PSM cannot be imposed by an invading army; or that ‘Afghans cannot be governed’; or that ‘A state cannot be built this fast.’ But there has to be some story to explain why PSM as an art (or craft or science or discipline or profession) should not have its reputation tarnished by the failure to build a capable state in Afghanistan. In short, knowing what will produce ‘better institutions’ where they are already pretty good (for example Singapore, Brazil, or even Indonesia) seems distinctly a second-order problem in a fragile state like Afghanistan.
Flailing states and isomorphic mimicry
The second area of incapacity for PSM is that there are many countries that have adopted all of the forms of well-governed states but in which ‘downstream’ implementation (downstream, that is, from the creation of a fundamental framework of state institutions) is riddled with both non-performance in the shape of absenteeism, lack of effort and so on, and corruption, with the result that that the operative functions of public sector institutions are lacking. India, where I have lived and worked, is a particularly striking example. Elsewhere I have labeled India a ‘flailing’ state (Pritchett, 2009). It can send a rocket to the moon. Yet many of its routine mass-level organizations are riddled with incompetence, lack of effort and corruption (Banerjee et al., 2008; Bertrand et al., 2007; Das and Hammer, 2007; Pritchett, 2012). The problem is that it is extremely difficult to link these ‘downstream’ deficits to the ‘upstream’ architecture of governance, because India does everything by the book in both its democracy (free and contested elections, freedom of speech, free press, institutionalized parties) and its institutions (independent judiciary, audit commissions, civil service pay commissions). That makes India a billion-person puzzle for PSM. It also means that it is hard to say what would make it better. The problem of ‘isomorphic mimicry’ (Pritchett et al., 2010) is that many countries look like they have adequate governance – they even get credit from development organizations for adopting international ‘best practice’ – while their public sector institutions have performed inadequately over many decades (Pritchett et al., 2010).
Solving such a problem is more difficult than building an institution or organization for the first time for two reasons. First, while it is expensive to build a new post office or university or health ministry on a greenfield site, organizationally speaking it is easy. However, once isomorphic mimicry sets in, those inside the organizations that are dysfunctional may deny that they are. In India, for example, the bureaucracy has created an artificial world of reports and data and words in which everything is going well. One of the interesting things about Banerjee et al.’s (2008) study of nurse attendance at clinics was that officially recorded absences were at very low levels. So, while only 30 percent of nurses were physically present, less than 5 percent were officially absent. The rest were in other administrative categories that were officially sanctioned and legitimate. Therefore any Indian government official could argue that absenteeism was negligible.
The second problem which isomorphic mimicry creates is that once organizations pass a tipping point where norms of behavior shift and non-performance is institutionalized, reversing that non-performance is much harder than creating a new organization. Once rents are ‘capitalized’ and have become the basis of settled expectations, reforms that upset the status quo are deeply resisted. In Indonesia in the post-Suharto era, the head of an infrastructure agency complained in the press, openly and using his real name: ‘I borrowed money to buy this post. If I cannot collect the usual kickbacks, how am I supposed to pay back my loans?’
The World Bank’s Approach document is remarkable and revealing in that it identifies the problem of isomorphic mimicry, and even discusses why it is hard to wean the World Bank off ‘best practice’ isomorphism. However, I am not convinced that it has found the solution. Where, then, might a solution be found?
Three possible alternatives
NGOs
NGOs have been a fashionable alternative since the 1980s. Bypassing the state using donors and philanthropists to finance NGOs as implementers does get the short-run job done, but even after almost 30 years of operation on a significant scale, they have yet to mature into state-like structures. Only in very few cases have they strengthened governance (Isham et al., 1995). ‘Development’ means being able to take a hot shower indoors because your local water company works – without having to going to ‘participatory’ meetings. It is not clear how this kind of bypass will get you back on the highway.
The private sector
A second alternative: could we just let the private sector rip, and the public sector wither, so that ultimately a new system of public sector capability (re)emerges around the de facto dominance of the private sector providers?
In some cases this is a terrific idea, and in some cases it is a terrible idea. For goods and services that are mostly private goods (in the economic sense of being rival and excludable) with few market failures, this ‘let it rip’ approach to the production of services (not their provision or regulation) is a great idea. For example, basic schooling has very few externalities, and so retreating into provision via performance-based regulation and equalizing finance while leaving the operational details of implementation to local producers is likely a good idea in low-capability states (Pritchett, 2013). But there are contrary cases. In one South Asian city with which I am familiar, the local urban politicians split the water supply into neighborhoods to create demand for their monopoly water tanker services. This was not an entrepreneurial response to public sector incapacity but the exact opposite – the public sector incapacity was induced in order to create monopoly profits for politicians masquerading as private entrepreneurs. Yet every efficient urban water system has a single set of pipes under the streets delivering clean, regular water because of fundamental technological cost characteristics. Displacement of incapable states by an inefficient private sector substitute is better than nothing, but it is not real progress.
Deep integration
The most promising alternative for improving state capability is deep and irreversible integration into an already capable system. The comparison between states in the former Soviet Union that joined the European Union (EU) and those that didn’t suggests that the very process of EU accession has built state capability. The same goes for the comparison between Puerto Rico, integrated into the United States, and its neighbor (Braun et al., 2004). However, deep integration is either available or not available (mostly it is not), and international agreements fostering economic integration, such as membership of the World Trade Organization, have not had the same benign effect on capability.
A proposal: problem-driven iterative adaptation (PDIA)
The third alternative is my preferred approach to building state capability, one that I have recently articulated elsewhere with colleagues (Andrews et al., 2012) and alone (Pritchett, 2012). The basic premise is that rather than functional success being built on creating the forms of public administration, or adoption of solutions that have proven successful elsewhere, effective organizations are built on success. Therefore the way to build organizational capability is, in the first place, to create success in solving problems. We postulate four principles:
Efforts should be problem driven, centering on locally identified problems. That might seem mere common sense, but most development focuses on transplanting solutions, working on problems identified by outsiders, such as the Millennium Development Goals, at an unrealistic pace and on an unrealistic scale. Finding solutions to these problems requires the authorization of positive deviancy. Units within the organization should be allowed the freedom to try things outside the existing practices. They should be accountable for their performance, not their conformity. Positive deviancy in this sense is justified where the problem that is being addressed is limited, urgent and measurable (failing which deviancy could as easily be negative as positive). That is, this is not an organization-wide reform but an ‘opt-in’ possibility. The turtle is a metaphor for this second principle. You cannot beat a turtle into moving forward. It has a hard shell, and it responds to external threats by pulling in its legs and waiting for the attacker to go away. Chronically dysfunctional organizations are like turtles: they just sit out external threats. The turtle needs an incentive to move; the threat of punishment is futile. Positive deviancy is effective to the extent that it is embedded in a process of learning that allows iterative adaptation, with incremental innovations being progressively refined based on feedback. We refer to this as a purposive ‘crawl of the design space’ using experiential learning (Pritchett et al., 2012). From another point of view, it is what Lindblom (1959) famously called ‘muddling through’. This builds organizational capability by feeding back into the space for ‘achievable practice’ within the organization as what is learned by the positive deviants is diffused through communities of practice. That is, the learning and scaling model is one of horizontal diffusion though existing networks within and across organizations rather than top-down imposition of ‘best practice’ policies.
If I had a magic wand which gave me the power to guarantee that my preferred reform would succeed, then my first choice for any given country would be a suite of effective public sector institutions and organizations – post office, police, courts, regulators, tax collectors – such as Denmark famously enjoys. Failing that, my second choice would be to transplant Denmark’s institutions and hope that they created a climate for effective organizations. However, if neither of those models turned out to be viable – and of course I wouldn’t be writing this if they had been in most places by now – then my third choice would be PDIA. It is, in terms of my earlier metaphor, a way of getting into the house by sneaking in the window. It is something one only does after trying and failing with all the keys. It might just work.