The policy implications of the Phillips curve relationship between inflation and unemployment have changed dramatically in the twenty-seven years since A.W. Phillips first identified a negative correlation between money wage changes and joblessness in Great Britain. Originally, Phillips' own findings suggested that policymakers could move the economy along his curve, trading off higher inflation for lower unemployment until the best (or least undesirable) attainable combination of both had been reached. Today, such a view is widely discredited. The statistical relation between inflation and unemployment has broken down and the Phillips curve is now generally viewed as offering no trade-off at all. This radical change in the policy implications of the Phillips curve did not occur all at once; rather it was the cumulative result of a series of theoretical innovations, which Thomas M. Humphrey chronicles in "The Evolution and Policy Implications of Phillips Curve Analysis." The two most important innovations were the natural rate hypothesis, which implies that unemployment can be reduced below its normal rate only by fooling the public with surprise inflation, and the rational expectations hypothesis, which implies that the public cannot be systematically fooled. Together, these two hypotheses imply that no systematic macroeconomic policy can affect unemployment. Even though no inflation-unemployment trade-off exists for policymakers to exploit, as Humphrey points out, policymakers can still contribute to reducing the variability and average level of unemployment by avoiding erratic policy changes and by enacting measures to improve the efficiency and performance of labor and product markets.