Delayed effects on migration intentions in an information provision experiment in Ghana

We report experimental results from Ghana, where treated subjects received information on regional income differentials. We do not see an effect on migration intentions directly after the intervention, but the effect of the treatment unfolds over time. Eighteen months later, subjects assigned to receive income information are on average significantly less likely to express enthusiasm for moving to another region, because individuals that had inaccurately high expectations about incomes elsewhere compared to their current place of residence are now more likely to want to forgo relocation. Contrary to common claims that effects observed in light-touch information experiments are likely to dissipate quickly, we suggest that some types of content in high-stakes domains such as migration can take time to reverberate and be incorporated into individuals' decision calculus. We also discuss that delayed effects may be uncommonly observed because long-term follow-ups are rare in the absence of short-term effects.