In a field experiment, we provide financial incentives to teachers framed either as gains, received at the end of the year, or as losses, in which teachers receive up-front bonuses that must be paid back if their students do not improve sufficiently. Pooling two waves of the experiment, loss-framed incentives improve math achievement by an estimated 0.124 standard deviations (σ), with large effects in the first wave and no effects in the second wave. Effects for gain-framed incentives are smaller and not statistically significant, approximately 0.051σ. We find suggestive evidence that the effects on teacher value added persist posttreatment. (JEL C93, I21, I28, J32, J45)
Stored only when both papers are in the corpus — 37 citations exist upstream.