1 Loeb and Page (1998) emphasize the difficulty of identifying salary effects, and argue that methodological problems lead to inconsistent estimates of the relationship between quality and pay. On the other side, Ballou and Podgursky (1997) provide evidence that school administrators do not hire the best teaching candidates, thereby weakening the link between quality and pay. Do higher salaries raise the quality of teaching? Many influential reports and proposals advocate substantial salary increases as a means of attracting and retaining more talented teachers in the public schools and of encouraging harder work by current teachers. Salary policies have also been cited as important for offsetting changes in demands outside of schools and for dealing with the potentially unattractive working conditions often identified in central city schools.The empirical evidence on the link between teacher quality and pay is, however, decidedly mixedraising doubts that there is a strong relationship between the two. Direct analyses of student achievement, for example, provide limited evidence of any systematic relationship. Two explanations have emerged in response to this evidence. On the one hand, some argue that the true relationship between teacher quality and salaries is quite strong, but methodological and data problems have impeded the identification of salary effects. Others take a less sanguine position, arguing that the evidence captures accurately the weak performance incentives in the public schools that lead administrators to make hiring and retention decisions that are not strongly linked with teacher quality. 1 The evidence is quite strong on one point: teacher quality is an important determinant of achievement (e.g., Rivkin, Hanushek, and Kain 1998).There are four main methodological problems that impede the estimation of the true relationship between teacher quality and salaries. Perhaps the most important is the difficulty of measuring teacher quality, because both direct measures of teacher characteristics and indirect measures based on student performance have proved problematic. Secondly, understanding the effects of salary policies requires distinguishing between shifts in salary schedules and movements along a given schedule (with, for example, increases in teacher experience), but this has proved generally difficult to do in past empirical work.