Comparison of international monetary policy measures

 Disclaimer: The data below are produced from the research of Leo Krippner and are not official Reserve Bank of New Zealand data. He will endeavour to update the data at the end of each month, but makes no guarantee on the timeliness of such updates. All data series are estimated and are therefore subject to changes upon re-estimations. These comparisons are also work in progress.

Estimated monetary policy measures

The International Shadow Short Rates file contains daily, month-end and monthly average Shadow Short Rate (SSR) estimates for the United States (US), euro area, Japan, and the United Kingdom (UK).

Download the International Shadow Short Rates (XLS 2.7MB)

The Measures of the stance of United States monetary policy page contains a longer series of US monetary policy measures. Note that those US SSR estimates are obtained from a different dataset and so are slightly different to the US SSRs on this page.

Measures of the stance of United States monetary policy page

The overview and documentation of how the estimates are produced is contained in the following section, and figure 1 plots the monthly averages.

Figure 1: Shadow Short Rate measures

Figure 1: Shadow Short Rate measures

Notes: The left panels of figure 1 contain the estimates over the entire sample. The right panels are from late 2007, therefore highlighting the lower-bound period. Major unconventional monetary policy events for the US are indicated as down arrows for easing and up arrows for tightening, and the progressive tapering of QE3 is indicated from the first tapering announcement to the announcement of the program’s termination.

Overview and documentation

For comparability, all of the estimates are obtained using the Krippner (2011-2015) shadow/lower bound framework with two factors, i.e. the K-ANSM(2), a fixed 12.5 basis point lower bound, and yield curve data with maturities from 0.25 to 30 years with the sample beginning in 1995. The first link at the end of this section provides further details and discusses how to interpret the different measures.

SSR estimates can be very sensitive to the model specification, data, and estimation method. The results presented here are designed to be as comparable as possible by holding each of those aspects consistent between applications. In addition, SSR results from different K-ANSM(2) applications are robust in profile, relatively robust in magnitude, and correspond well with unconventional monetary policy events. These properties do not generally hold for SSR estimates from three-factor models, which includes K-ANSM(3) models and the Wu and Xia (2016) model. The second link below and the historical documentation in the following section contains further details.

The SSR estimates are all nominal, and so accounting for changes in inflation expectations would be required to gauge changes in the real stance of policy. The profile of the Japanese results in particular will be influenced by this aspect, because the inflation target and core inflation have both increased since early 2013.

Download the documentation for United States measures of monetary policy May 2016 (PDF 239KB)

Download a comment on Wu and Xia (2015), and the case for two-factor Shadow Short Rates (PDF 815KB)

Historical estimates and documentation

The historical United States monetary policy measures file contains the historical estimates, for reference, before improvements were made to the model in May 2016. An overview of the changes and their impact are available in appendix B of the May 2016 documentation.

Download the historical United States monetary policy measures (XLSM 3.4MB)

The September 2014 documentation details the original model.

Download the documentation for United States measures of monetary policy September 2014 (PDF 611KB)

Code for lower-bound term structure modeling

The MatLab code page contains the original MatLab code and the new code used from May 2016.

MatLab code page

The Python code page contains the Python code equivalent to the original MatLab code. It was kindly translated and made available by Amandeep Singh.

Python code page