Incredible policy attention has been given to the claim that an increase in the quality of education as measuredby international tests (e.g. PISA tests) has a significant impact on the GDP long-run growth rate (Hanushek and Woessmann, 2015). This study is based mostly on aggregate data from the second half of the century, and never addresses the question of the current paper, which is whether the impact of the quality of cognitive skills affects the level of GDP per capita or the long run growth rate. Focusing on this question, we construct a variant standard growth model in which cognitive skills have theoretically both a level and growth rate effects by assumption. Estimating this model using standard cross-country data and panel data, cognitive skills measured by the methodology of Hanushek and Woessmann (2015) have a significant level effect on GDP but not a growth effect. Therefore, the cognitive skills improvement impact economic growth is bounded.