The Bank of Japan and Breakevens

Scott Sumner (via Matt Yglesias) has a very impressive graph of breakeven inflation rates for yen denominated vs inflation protected securities in Japan (which seems to have something to do with Fathom Consulting just to cite).  It shows a marked increase from around 0% to around 0,8% from early February 2011 to  late April 2011.  The BOJ announced asset purchases and a 1% inflation target February 2011 which shows up on google indexed sites reported on February 14 (they sent Scott Sumner a Valentine).   There is a further increase from 0.8% to 1.2% following the announcement of a 2% target January 22nd.

I am impressed.

I have three thoughts.

1) given standard assumptions (that is the efficient markets hypothesis which I’ve never believed for a second in my life) the effect of an announcement on asset prices should be instant.  The longest interval which is usually considered in event studies is a day.  Asset prices are just not supposed to change gradually in a predictable direction over a period of months.  This point doesn’t interest me, because I think the EMH is nonsense.  It is true IIRC that, aside from this debate, graphs with months on the X axis just aren’t used to study the effect of policy shifts on asset prices.

2) the recent policy shift clearly includes a major change in the fiscal authority (a new Diet with a new majority).  Events following the election can’t be use to distinguish the effects of fiscal and monetary policy or show that either one would be effective in the absence of the other.

3) I have noticed (to my amazement) a strange correlation between US TIPS breakevens and inflation over the preceding year.  This makes no sense (actual inflation over the duration of the breakeven is not nearly so highly correlated with the past years inflation).  This is relevant to the question of what monetary authorities can do, because last years inflation if a moving average of price increases over a whole year so monetary authorities can’t make it change quickly.  Also it is widely agreed that most prices are sticky so again,  a change in inflation soon after a shift in monetary policy is not plausibly due to the shift (“long and variable lags” — Milton Friedman).

From January 2012 to April 2012 the past year’s CPI inflation rate increased from 0.1% to 0.5%.  Now for the USA I get a regression coefficient of the breakeven on the past year’s inflation less than one (and lower for past CPI inflation than for core inflation).   This time I would need a regression coefficient of around 2.

Then in May the past year’s inflation fell to 0.2% and the breakeven fell to 0.6%.  That’s a lot more like the US pattern.  This leaves a substantial increase in the breakeven.

A very silly graph

A slightly less silly graph