Innovation plays a pivotal role in fostering economic growth, yet there is a limited understanding of whether it can be taught. I conduct a randomized evaluation of an education program implemented by a state government and a nonprofit organization, providing an opportunity to 6,224 8th-grade students from disadvantaged backgrounds to develop frugal innovations for global and local problems. To assess students’ innovative ability, I created a novel scale with inputs from experienced inventors and used a lab-in-the-field game from experimental economics. The program consistently has a positive impact on students’ innovative ability, as demonstrated by improvements on both the novel scale (0.20 standard deviations) and in a lab-in-the-field game (0.12 standard deviations). Notably, the ability for innovation is not correlated with academic achievement or IQ. On the contrary, the gains in the ability for innovation came at the expense of interest and performance in math, declining by 0.30 standard deviations and 0.13 standard deviations, respectively. These results have significant implications for governments around the world in their capacity to impart an important higher-order skill in schools that strongly correlates with long-term economic growth.