Archive for the ‘e488-phillipscurve’ Category

A Response to Lansing’s “Can the Phillips Curve Help Forecast Inflation?”

Monday, January 28th, 2008

Rather than look at articles that delve into the time line behind the development of the theory, I thought that it would be useful to explore a practical application of the Phillips curve.  To me, the application of a theory and/or model (once you understand its background and assumptions about the world) is more interesting and useful than a simple discussion of the evolution of a particular way of thought.  Thus, I began exploring websites that looked at practical uses for the Phillips Curve and came across Kevin J. Lansing’s FRBSF Economic Letter“Can the Phillips Cure Help Forecast Inflation?”  Lansing starts the brief by giving the reader a basis for the question as to whether the Phillips Curve is a good forecasting tool for future inflation measures.  He notes that in the 1960s many economists and policymakers believed there to be a “stable trade-off” between inflation levels and levels of the unemployment.

 The substance for the forecasting application of this model really comes into play during the 1970s when Edmund Phelps (1967) and Milton Friedman (1968) refuted those earlier claims–as well as A. W. Phillips (1958) findings that policymakers could lower the unemployment rate by creating higher inflation–and stated that this “stable trade-off” could only be realized in the short run.  Phelps and Friedman’s comments led to a short-run Phillips curve, that is, the inflation and unemployment levels tend to move in opposite directions in the short run.  Interesting enough, the short-run unemployment that the Phelps-Friedman argument refers to is more commonly known as the Non-Accelerating Inflation Rate of Unemployment (NAIRU), which is the level of unemployment that holds the inflation rate at a particular level.

This short-run Phillips curve model became an interesting topic for Atkeson and Ohanian (2001) and Fisher, Liu, and Zhou (2002) in which those papers looked at the model as a useful tool for forecasting inflation rates.  Atkeson and Ohanian (2001) divide their sample into two periods: 1960-1983 and post-1983.  The relationship in the first period is negative, but in the second period, the relationship produces a line that is much flatter, indicating even less of a relationship between inflation and unemployment.  Fisher, Liu, and Zhou (2002) follow-up the 2001 study and examine closely the post-1983 period.  The paper concludes that the Phillips curve model can correctly predict the direction of the rate of inflation 60-70% of the time.

The final reason for Lansing discussing the application for the Phillips curve as a useful forecasting tool is because of the relationship between unemployment and inflation in the 1990s.  As the traditional model shows, there is a trade-off, or inverse relationship, between inflation and unemployment.  Nevertheless, during this decade, inflation and unemployment both moved in the same direction–downward.  Lansing gives reasons for this observed trend–inflation and unemployment remained low due to the increase in the growth rate of worker productivity due to the creation of new technologies.  Thus, due to the fact that multiple regression lines could fit the wide dispersion of data–suggesting that the inflation-unemployment relationship isn’t precisely shown with the short-run Phillips curve–there is evidence indicating that the short-run Phillips curve isn’t useful in forecasting the magnitude of future inflation, but perhaps is better at forecasting the direction of the change of future inflation.

Source: Lansing, Kevin J.  2002.  Can the Phillips curve help forecast inflation?  FRBSF Economic Letter 02-29 (October 4): 1-4, (accessed January 28, 2008).