Negative Interest Rates and a Term Structure Puzzle

James Hamilton provided us with another interesting discussion on negative interest rates:

we now have several years of experience from Sweden, Denmark, Switzerland, Japan, and the European Central Bank in which the central bank successfully induced negative interest rates in hopes of stimulating a greater level of spending on goods and services.

Please read the entire post including some interesting comments. Alas I must be late to the party as I could not provide a reply to an interesting query from Barkley Rosser:

Does anybody have an explanation as to why when a nation has negative nominal target interest rates it often seems that the time horizon of government securities that end up having the most negative rates are often at the two years time horizon? Look at the Sweden case, where this has been the case, and it has also been in quite a few other nations as well. I have yet to see an explanation of this peculiarly non-monotonic yield curve in this situation so often.

Maybe Europe has turned Japanese. I've been looking at an Excel file of their government bond rates provided by the Ministry of Finance (not the Bank of Japan). Japan had low but not negative interest rates before 2012 with a somewhat upward sloping term structure. Since 2012, two features describe this data: (1) one-year rates hovering around zero - sometimes positive and sometimes negative; (2) two-year rates hovering near the one-year rate and it times just below them. What is driving this? I have no answer.