As you can see in Figure 17-4, the impact of education is larger with FE than with OLS. An additional year of education increases wages by 4 percent with the OLS estimates and increases wages by 8.7 percent with the FE estimates (I discuss the interpretation of coefficients with the
log-linear specification in Chapter 8). Consequently, the OLS results underestimate the impact of education by ignoring unobserved individual heterogeneity.