In an artefactual field experiment in rural Namiba, we examine the effect of a psychological self-affirmation intervention on productivity among poor and marginalized groups. Poverty and the associated stigma as well as the exposure to ethnic discrimination among marginalized groups potentially constrain productivity through cognitive taxes or by lowering one's perceived ability. A self-affirmation intervention that has been successful in mitigating the impact of these threats in previous research by improving self-image is thus hypothesized to also enhance productivity in our context.

As part of the self-affirmation intervention, participants are asked, during a verbal interview, to recall personal situations in which they felt successful and proud. In a control condition, the interviews dealt with participants' daily routines. Afterwards, we offered participants to work on locally common job for a local farmer by cracking peanuts for one hour under a piece-rate payment scheme.

We find a surprising result among self-affirmed participants. Contrary to previous evidence and our hypothesis, productivity in the self-affirmation treatment is significantly lower compared to the control treatment by a magnitude of about 15%. Through coding of the interviews by two independent raters, we show that the negative treatment effect is driven by participants who have difficulties answering the self-affirmation question. We further show that this effect is not endogenously driven by an underlying ability influencing both productivity and the ease of recalling a success story: We exploit the correlation of participants' (exogenous) level of education and find that participants with a lower level of education are less productive after the self-affirmation intervention but not in the neutral condition. We consequently conclude that being unable to recall personal stories of pride and success either taxes participants' cognitive resources or lowers their perceived abilities and thereby impairs their productivity.

Our results provide valuable insights for careful policy design as they demonstrate a case in which a self-affirmation intervention backfires and proves to be detrimental for those who would need it most: the poor and marginalized.