# How do I verify Philips curve empirically for recent years?

I wanted to verify the relationship between inflation and unemployment empirically, so I took data from the World Bank and used excel to correlate them.

I got a very, very bad correlation between unemployment and inflation (from 1991 to 2016). Am I doing something wrong here?

I took these two sets of indicators: (1) Unemployment, total (modeled ILO estimate) and (2) Inflation, consumer prices. Here are the links: http://data.worldbank.org/indicator/SL.UEM.TOTL.ZS and http://data.worldbank.org/indicator/FP.CPI.TOTL.ZG

I downloaded the data from here and I studied the data for the United States of America. Is the low correlation because of the fact that the market needs time to react to unemployment rates? If so how did Phillips study it in the first place?

• I looked at the site quickly, are the data only available as annual data? A first step would be get monthly data. Annual data misses a lot of information around recessions. For the United States, FRED (St. Louis Federal Reserve) provides such data for free. Commented Apr 21, 2017 at 13:23
• You don't. Well...that is my guess, at least. This recent period is characterized by a binding ZLB. The economy behaves quite differently at ZLB. The central issue is that the monetary authority can't adhere to the Taylor principle and so real interest rates act differently. Anyway...at least consider unemployment adjusted for changes in labor force participation rates. Then maybe it won't be so egregious.
– 123
Commented Apr 21, 2017 at 13:38

Simple correlation is not a good approach to evaluate the Phillips curve, because you are mixing many periods together. To get a better grasp of what this relationship might have been in the past, use a graphical approach.

I suggest you to do something like this:

So basically, you plot the points, but add a line that connects them temporally. This might give you a better idea of some dynamics. It is clear from that graph that a simple correlation cannot capture the complex dynamics of inflation and unemployment.

For example, this is an animation of the evolution of the Phillip curve for the US:

The author identifies three or four PC over 40 years, reflecting different political/economic periods. A simple correlation analysis however would show none.

Source of images: here.