Turkey changed its compulsory attendance law in the 1997-1998 academic year The requirement increased from five to eight years of primary education. After the law change, there is an abrupt increase in the high school enrollment rate. This is despite the fact that the law does not cover high school education. By using the 2003 household budget survey and differences in differences methodology we find that a typical student is 3.2 percent more likely to have high school education if he is subject to the new law Moreover the largest increase in the likelihood of having high school education is attained by the students who have the weakest socioeconomic background. These findings are consistent with the standard sorting model but contrast the standard human capital model.