Wrapping up: Thinking about evidence

Many anti-corruption researchers and practitioners will be familiar with two recent attempts at mapping the evidence on the effectiveness of anti-corruption interventions (e.g. Rocha Menocal et al., 2015 and Zaum et al., 2012). While both started out with the aim of establishing ‘what works’, in the end both reports tell us more about weaknesses within the evidence base: in short, we actually know relatively little about whether any given anti-corruption intervention is likely to work. Rocha Menocal et. al. (2015), for example, looked at twelve intervention categories but found only two—decentralisation and social accountability mechanisms—with a relatively large enough body of evidence to make any claims on efficacy; however, even within those bodies of evidence, whether or not the evidence tells us ‘what works’ was categorised as  ‘contested’ by the mapping report’s authors.

More/better evidence is needed to tell us what works, but not enough attention has been paid to potential unintended consequences of anti-corruption interventions. A growing body of evidence is helping to demonstrate why testing interventions, even those that seem positive and/or benign, is important.

Almost all anti-corruption strategies have an awareness raising component, whether this is through campaigns, billboards, embedding messages into school curriculum or so on. Despite widespread use, until 2016, the influence of awareness raising activities hadn’t really been scrutinised. Since then, five studies have examined what happens in a variety of countries and contexts when ordinary citizens are exposed to different anti-corruption messages (Corbacho et al., 2016; Peiffer, 2017 and 2018; Kobis et al., 2019; Peiffer and Walton, 2019; Cheeseman and Peiffer 2020).

Worryingly, the emerging consensus is that anti-corruption messages do not work; none of these studies show that anti-corruption messaging prompts the intended outcomes: discouraging corrupt behaviour and encouraging citizens to report/fight corruption. Worse still, there is growing evidence that through priming citizens to think about corruption in their country, anti-corruption messages may actually be backfiring, where ‘exposure to anti-corruption messages fails to discourage corrupt behaviour and, in some cases, actually makes individuals more willing to pay a bribe’ (Cheeseman and Peiffer 2019: 8).

Prior to these studies, few concerns had been raised about anti-corruption awareness-raising efforts potentially causing harm, let alone questioning whether or not there’s sufficient evidence that they actually work. Given the ways in which various actors use the public’s frustration with corruption for their own agendas, being clear that interventions aren’t causing harm, and understanding the relationships and interactions we confront, is just as important as betting understanding what works. And to do this, we need evidence.

In 2018, we wrote: ‘Without…engaging in the political dynamics that underscore corruption, anti-corruption efforts will continue to fail. Thus, anti-corruption reformers should ask themselves: What functions does corruption serve as well as what obstacles exist to eradicating it? Arguably, without answering the first question, one cannot begin to adequately answer the second’ (Marquette & Peiffer, 2018: 509-510). This may mean turning to indirect approaches that won’t necessarily directly target the corrupt behaviour. It may, in fact, mean not fighting corruption at all, instead targeting the necessity, perceived or otherwise, to engage in corruption. Many of the examples we provide in this Framework would certainly be best approached indirectly, possibly without even using the so-called ‘C word’ (Wolfensohn, 1999).[8]

Like with crime, terrorism, sexual exploitation and so on, however, sometimes direct approaches are exactly the right way to work. Just like we wouldn’t expect the police to turn a blind eye to drugs trafficking or the online abuse of children until underlying social, economic and political conditions can fully be addressed, we wouldn’t expect a blind eye to be turned to politicians giving out large procurement contracts to their cronies or to health workers denying vulnerable patients treatment because they can’t afford to pay bribes.

Hindess (2008: 22,25) offers an important caveat about indirect approaches, which he argues risk interventions that only target areas that are ‘politically unproblematic’, shifting ‘the focus away from the issue of dealing with bad individuals and/or bad practices to the very different issue of societal reform—in this case of changing the social context in which such individuals and/or practices are able to flourish’. Such a focus, he argues, means that vested interests and their power are often left unchallenged, and angry citizens are left feeling as if there’s one rule for the powerful and one rule for everyone else.

We don’t yet have a strong enough evidence base to say with confidence that indirect approaches are more effective than direct ones, but understanding the functions that corruption fills is likely to be essential for helping to decide when direct approaches are right; when indirect approaches are better, more effective and safer; or – most likely – how to more effectively combine the two. Again, this isn’t about ‘sequencing’, where we only focus on fixing the underlying problems first; the social, economic and political costs of many forms of corruption are often too high to wait for this. But without actually fixing these problems at some point, we’re very likely to be stuck in this same anti-corruption loop without any end in sight.