Given the amount of time and effort spent analysing corruption in recent years, it should come as no surprise that there are a wide variety of potential remedies on the market. Indeed, the world is most certainly not suffering from a dearth of toolkits, action plans, agreements, conventions, treaties and agendas for change. The aim of this chapter and the two that follow is not to unpack every anti-corruption option available, but rather to look at the evidence for what seems to work and what does not. The latter point is particularly important. Despite the multiple calls to arms of recent years, the evidence that genuine progress has been made is disappointingly thin on the ground. There is subsequently a case for both rethinking what we think might work and assessing what indeed “success” might mean.

For starters, any anti-corruption attempt that claims to want to eradicate corruption or adopt a “zero-tolerance approach” should be treated with more than a dose of scepticism. The aims are too grand and, quite frankly, incompatible with the rough and tumble of everyday life. Corruption will often be deeply embedded in given social and political practices, and part of complex and ever-changing power relationships. This is all before, of course, any consensus has been reached as to what exactly the problem is, and what needs to be done to put it right. The specific aims of any attempt to tackle corruption, in other words, need to be carefully defined.

Successful anti-corruption is, by definition, a piecemeal affair. As Michael Johnston has noted, reforms often take a long time to work; they need a modicum of good fortune along the way and can be “a lot more messy and acrimonious” than is generally anticipated (Johnston 2013, 2016: 16). One-size-fits-all policies, even of the most obvious nature, can be counterproductive. A given anti-corruption mechanism may work well in one setting but have little impact in another, whereas it could cause far more problems than it solves in a third context. Carefully designed strategies focused on local contexts and specific problems are very much the order of the day, even if they, too, fail from time to time.