Saturday, October 17, 2009

Recalculation in Action


Tail Trimming the Inflation Distribution

Macroblog has a great post on the relative distribution of inflation over the array of consumer goods used by the BLS. He uses the trimming method.

Let me try and walk through it. There is a distribution of consumer goods in the inflation basket. We would expect that the array of consumer goods smoothly and symmetrically distributes price changes during stable periods. Some of these change price fast, some slow; the CPI number reported is the average. When trimming they remove both the good that changes the least and one that changed the most, the goods at the both extremes of price changes. So, when we start removing the goods at the extremes ends of the inflation distribution, we would want the average inflation of the remainder to be stable, if inflation is stable.

The graph above takes the inflation data for the few months previous, but the X axis, shows the inflation after some percentage of the extremes have been removed. As more of the extremes are removed, the remaining inflation stabilizes, the bulk of consumer goods having a common level of inflation.

What I want to point out is that since Jan, the trimmed inflation series have become more stable; there are fewer extremes in the data. We see this if the data series smoothly stabilizes as the line moves right. Earlier series show more rapid excursions than later, implying that we are Recalculating the best way to use the scarce resources.

No comments: