Friday, March 13, 2015

The 3 dimensional Phillips curve as a forecasting tool

 - by New Deal democrat

Why do I continue to harp on the Phillips curve being best viewed as 3 dimensional, with the third dimension being basic commodity prices?  Because it helps forecast where the traditional, 2 dimensional Phillips curve is likely to be in a year or so. And since the Fed is focused on the "non-inflationary rate of unemployment," we at very least want to maximize employment and wages without causing self-defeating inflation.  Which means we don't want the Fed to apply the brakes too early.

I can actually show you this very neatly in traditional, two-dimensional graphs.  Some might object that I am "double-counting" inflation, since one axis is commodity prices, the other consumer prices. But commodity prices do not move in sync with consumer prices. In the graph below, both commodity and headline consumer prices are normed to 100 in 1982:

You can see that over the long term, commodity prices have gone up at less of a rate than consumer prices, but with significant exceptions in the early 1970s and the 2000s.

Now let's take ratio of commodity prices to consumer prices, and compare it with the "misery index" which is the addition of the two components of the Phillips curve, the unemployment rate + the YoY consumer inflation rate:

With just a few exceptions, (the late 1960s and 1989-90), for the last half a century the two have moved in tandem.  This is strong evidence that the relative strength of commodity vs. consumer prices does indeed move the Phillips curve along a third axis.  Furthermore, note that the "misery index" lags the ratio of commodity to consumer prices. By knowing the relative YoY% changes in commodities vs. consumer prices, we can forecast where the Phillips curve will be about 12 months later.

And to be clear, here is the YoY% change in commodity prices (blue) vs. core inflation (red, first graph) and the unemployment rate (red, second graph):

Commodity prices (the Z axis on the 3 dimensional Phillips curve) lead both core inflation and the unemployment rate.

Finally, here is a scatterplot of commodity prices and the unemployment rate over the last 5 years:

The values have been consistently shifting to the lower left, with the last two months at the extreme bottom left. This tells us that the 2-dimensional Phillips curve has also been shifting downward and to the left.

In short, treating the Phillips curve as 3 dimensional gives us the ability to forecast the values of the traditional two dimensional Phillips curve 1 year out. In 2015 this means the traditional Phillips curve should shift to downward and to the left of where it has been in the last few years. Since the values of the last several years already indicated an overshoot of the Fed's inflation target of less than 1%, the 3 dimensional Phillips curve forecasts that unemployment can fall well below the Fed's range of 5.2%-5.5% without triggering core inflation in excess of 2%.