Do behavioural anti-corruption interventions work? New review looks at the evidence
A new paper by our Public Governance team explores the evidence behind behavioural anti-corruption interventions.
Donors, governments and anti-corruption practitioners seeking alternative tools to address systemic corruption are increasingly turning to behavioural science. Behavioural anti-corruption approaches appear promising because they respond to a growing body of descriptive evidence on how certain social norms and mental models drive corruption, particularly in fragile contexts. Interventions that target social norms and seek to shift people’s behaviours away from corrupt practices could be more effective and long-lasting than ones that, for example, simply add more regulations and controls.
Yet few large-scale anti-corruption programmes have so far been informed by behavioural insights – in part due to a lack of evidence on where such an approach would be appropriate, what works and what doesn’t.
That evidence is slowly becoming available, thanks to an increase in the past five years in what can be called Social Norms and Behaviour Change (SNBC) intervention studies. Many have yielded positive effects and demonstrate the potential of SNBC interventions to tackle systemic corruption, but some studies have encountered counterproductive effects of anti-corruption messaging.
Based on a synthesis of the evidence, the paper summarises a set of behavioural explanations (i.e. insights and pitfalls) for why some of these SNBC approaches have failed, while others have been effective. The aim is to provide practitioners designing SNBC interventions with evidence to help them develop effective programmes and avoid common pitfalls.
Corruption risks and health crises in Malawi: new GI-ACE research project
We are excited to launch a new project on addressing corruption risks in the emergency response…
Culture and corruption: a complex relationship
If someone gives a healthcare worker a “gift” in return for faster treatment, is that a bribe or…