In the 1990s and 2000s, unemployment fell while inflation did not rise, which contradicts the theory based on the Phillips curve. Economists attribute the deviation from the theory to the use of supply-side policies that increased overall productivity and lowered the natural unemployment rate. Examples of supply-side policies that increase productivity include lowering corporate tax rates, reducing regulation, improving education and job training, reduced power of trade unions, and adopting performance-based pay. These policies may help economies to sustain growth while maintaining full employment and low inflation, but they can also be slow and costly to implement.
For example, during 2006–09, housing prices in the United States fell by one-third. This led to a reduction in household wealth and difficulties for financial institutions. The end result was a large decline in aggregate demand and a steep increase in unemployment. From 2007–10, the decline in aggregate demand increased the unemployment rate from below five percent to about ten percent and reduced the rate of inflation from about three percent to one percent. This is an example of an economy moving down a short-run Phillips curve. However during 2010–15 as the economy slowly recovered and the unemployment rate fell back to its natural rate of five percent, the inflation rate remained between one and two percent. A common explanation for lower unemployment and stable inflation is the commitment by the Federal Reserve to keep inflation at two percent and to keep interest rates low for several years.
Many economists question the usefulness of the Phillips curve because of its lack of predictive power. Originally it appeared that the Phillips curve showed a one-to-one relationship between the unemployment rate and inflation rate, and it was thought that the inflation rate could be set by controlling unemployment. However, even modified versions of the Phillips curve that try to consider expectations fail to predict interest rates; since the 1970s, the Phillips curve has proven to be, at best, inconsistent at predicting the relationship between unemployment and inflation. Though it may lack the ability to quantitatively predict unemployment and inflation, many economists still believe that the Phillips curve is useful for explaining and understanding how they are related.