My latest is to estimates some Phillips curves. From the exercise, I learn that Mark Sadowski knows more about the macro-monetary policy literature than I do (OK I knew that -- I focus on oh growth or something or maybe robust econometrics). Nick Rowe also suggested that the results are just what one expects if the FOMC successfully targets the inflation rate (this is true). I replied at length in comments and copy and paste my reply here
April 13, 2014 7:14 am
This sort of result is exactly what you should expect. if the central bank is targeting inflation. The whole point of inflation targeting is to ensure that expected inflation does not vary over time, so you should not be able to estimate any effect of expected inflation on actual inflation.
Robert: Nick yes. The results are consistent with what central bankers call "anchored inflation expectations" and would occur if inflation were successfully targeted. There are two problems. First inflation has not been successfully targeted, inflation has not been equal to the target -- this could be because the true target is 0 to 2% not 2 % (or 2-4% as a goal not a target in the 80s). Second TIPS breakevens *are* still correlated with lagged inflation. There are two facts (both reported at this blog) one is that lagged inflation has a negative correlation with wage inflation, the other is that the R squared od TIPS breakevens on lagged inflation is about 50%.
You can explain the styllized fact which I just reported, but not the one I reported February 25 2013
Also there are survey's of price level forecasts made by people chosen as experts. I haven't reported much on this, but I have been looking at median CPI forecasts from the Livingston survey
Here I note that Volcker seems to have had significantly less inflation fighting credibility than Arthur Burns -- a very standard argument for how actual expectations differ from adaptive expectations is in total contrast with the data.
Here I note that decade averages Livinston median forecast errors correspond almost exactly to decade averages of unemployment.
All are evidence that very crude models of expectations work very well -- when modelling expectations. Now the Livingston team changed the survey in 2003 and I haven't dowloaded data from post 2003 surveys. My excuse is that I am keeping them for out of sample forecasts and giving myself no chance to data snoop. The real reason is that I am lazy. So I only have one year ahead forecasts of inflation up to 2004 (forecasts made in 2003).
Here is a a scatter of lagged annual personal consumption deflator excluding food and energy inflation and the annual CPI inflation forecast calculated from the mediian Livinston CPI forecast. I look at data since 1990 (recall the expecations unaugmented Phillips curve fits the data 1990-2013 well). There is no sign from the Livingston survey that inflation expectations are anchored.
Between the TIPS results (for recent years for which FRED has TIPS data) and the Livinston survey results (for older years before they changed the survey) I am willin to assert that inflation expectations are not anchored at all -- that TIPS purchasers and Livinston survey participants do not believe that the FOMC is successfully targeting inflation year by year (actually the TIPS breakevens are 5 year breakevens).
I should add that the median Livingston participant makes extremely anchored forecasts of inflation over the long term (10 years IIRC) forecasting around 2.5% (IIRC) in survey after survey.
Not ready for prime time, but I am convinced that the key issue is downward nominal rigidity not anchored inflation expectations.