Shadow rates at the zero-lower bound

Understanding the impact of unconventional monetary policy is tricky because its much more difficult to assess the actual stance of monetary policy when short term interest rates are at the zero lower bound.

One way around this problem is to calculate a "shadow rate", which uses information from the yield curve that isn't bound at zero to determine where short term rates might be if they could go negative.

A nice overview of these ideas can be found in this presentation by James Bullard, president of the FRB St. Louis.  He presents quite a bit of research from Leo Krippner, a researcher at the Reserve Bank of New Zealand, of which a summary can be found here.  

Using similar ideas are Wu and Xia, who have published a very wonky academic working paper found here.  They have provided their estimates of the shadow rate for researchers to use in other models. They have used a different estimation technique than Krippner, although the idea is similar.

In their paper, they use this shadow rate in a vector autoregression model to measure the impacts of policy, which has been used by researchers prior to the financial crisis when policy rates were positive. This suggests unconventional policy has lower the unemployment rate by ~25bps from the counterfactual.

What is interesting about the Wu and Xia estimates is that shadow rates have become increasing negative in the past few months (data is to December).  This is despite tapering bond purchases and an increase in 10 year treasuries into year end.

But is policy really easier? A theoretical shadow rate is interesting and is useful in modelling, but how do we think about it in terms of the real economy? Krippner shows in his paper that this has been useful in explaining movements in the NZD, which is critical to a small open economy.  But for the US, higher borrowing rates for households and businesses is what matter, not a theoretical short-term rate. So I wouldn't be reading too much into month-to-month moves.