A nonparametric estimation of the Bank of Canada's reaction function
This dissertation uses the Kernel method to estimate a nonparametric forward-looking monetary policy reaction function for the Bank of Canada using real-time data. The purpose of using a nonparametric method is to leave the functional form of the conditional mean function unspecified, allowing the data to determine the nature of the relationship between the dependent and explanatory variables. In addition, the Kernel method does not out of necessity exclude the possible presence of asymmetry in the BoC reaction function. Baseline estimates are conducted using OLS and the GMM methods. Subsequently, the Kernel method is applied to estimate a nonparametric reaction function along a number of inflation forecast horizons. Results indicate that the Bank of Canada behaves in a forward-looking manner, targeting inflation about two quarters ahead. A number of tests are performed to check if the BoC acts in an asymmetric manner and evidence suggests that the Bank acts more forcefully to positive deviations of inflation from target than to negative deviations. Similarly, the BoC acts more vehemently to positive output gaps than to negative output gaps.
Khan, Farrukh J, "A nonparametric estimation of the Bank of Canada's reaction function" (2011). ETD Collection for Fordham University. AAI3466709.