Wednesday, April 23, 2014

Data and Japan

Steve Williamson has a new post on monetary policy.  He argues that the data supports his view that quantitative easing lowers the rate of price inflation and low interest rates cause low inflation.  Unfortunately, he made a pretty significant error in constructing the data, and after fixing the error I think there is reasonable evidence that he's wrong.  [As a Ph.D. student I live in constant fear of using a data set incorrectly, so it's nice to see that even high ranked tenured professors can make mistakes.]

Here is the graph he shows:

The graph is pretty striking.  So much so that I started googling for stories about Japanese deflation.  The graph shows year over year measures, so the price level has to drop about .9% in a month to get from 1.5% in February to .4% in.  That would be an annualized inflation rate of -11%.

It turns out that the March data doesn't exist yet.  From the comments, it sounds like Stephen conflated two series, the Ku-area of Tokyo CPI and the Japan area CPI.  Specifically, the way he did it created a large error, because he was using the price level of Japan CPI and inserted the price level of the preliminary March Ku-area CPI into it.  These two series are completely different and use different bases.  On his chart it looks like the price level (2010 base) fell from 100.7 to 99.8.  In reality, the Ku-area price level rose from 99.3 to 99.8.  The preliminary results show more inflation, not less.  Below are the two graphs:

We should know by April 30th what happened to Japan CPI in March.  If the Japan numbers are up as the preliminary Tokyo numbers are, then Japan CPI will be getting pretty close to the 2% target.

Stephen Williamson's model predicts in the long run that low rates and Quantitative Easing should cause low inflation.  That isn't evident here and it isn't evident in the breakeven inflation rates for Japan.

This graph shows a quick and dirty market measure of inflation over the next five years.  It shows 2.3% and has been rising.

So we have Sweden that raised rates and then deflation followed and Japan that has not raised rates and not had QE and so far has seen measured inflation and expected inflation rise.  My prediction is that if Japan inflation falls a lot, it will be because Japan raised rates or stopped QE to early and caused expected inflation to fall dramatically long before measured inflation falls.  

No comments:

Post a Comment