This is valuable analysis, although with a limited amount of data. Two comments ob my side. 1. The PPI inflation increased earlier in the last inflation episode, and more importantly than CPI. Keep in mind the limited number of data points, one strategy would be to replace CPI with PPI. 2. Would you have found similar results for the previous hiking cycle in 2004? Well done and looking forward to the upcoming analyses.
Anjeza, thank you. With so few data points, there is little that can be done. But the idea of looking at the PPI is sound and I will do so.
The problem with going back to pre-2015 data is that r* was most likely different and gradually changing over time. It is not obvious how to control for that.
Presumably a reaction function estimated over a longer period would include the EURCHF exchange rate, or even a fiscal constraint (I confess I'm not familiar with the literature on the SNB RF).
Spyros, thank you. This is a very good point. But, as you say, to estimate additional parameters I need a longer sample. In this version the effect of the exchange rate is captured through its effect on inflation. Including the exchange rate as a regressor is a problematic because of reverse causality: expectations of future interest rate changes will impact on the exchange rate already now so instrumental variables must be used.
Dear Stefan,
This is valuable analysis, although with a limited amount of data. Two comments ob my side. 1. The PPI inflation increased earlier in the last inflation episode, and more importantly than CPI. Keep in mind the limited number of data points, one strategy would be to replace CPI with PPI. 2. Would you have found similar results for the previous hiking cycle in 2004? Well done and looking forward to the upcoming analyses.
Anjeza, thank you. With so few data points, there is little that can be done. But the idea of looking at the PPI is sound and I will do so.
The problem with going back to pre-2015 data is that r* was most likely different and gradually changing over time. It is not obvious how to control for that.
Presumably a reaction function estimated over a longer period would include the EURCHF exchange rate, or even a fiscal constraint (I confess I'm not familiar with the literature on the SNB RF).
Spyros, thank you. This is a very good point. But, as you say, to estimate additional parameters I need a longer sample. In this version the effect of the exchange rate is captured through its effect on inflation. Including the exchange rate as a regressor is a problematic because of reverse causality: expectations of future interest rate changes will impact on the exchange rate already now so instrumental variables must be used.