Dark Corners.JPG

Computing Better Paths for the Policy Rate than Backward-Looking Taylor Rules

Using a stylized model of the U.S. economy, this course will show participants how to compute much better paths for the Fed Funds Rate than backward-looking Taylor rules.  These paths are based on minimizing quadratic-loss functions and are much more consistent with the Fed’s behaviour and the FOMC’s interpretation of their dual mandate of low inflation and maximum employment.

Possible Costs of Overheating and Boom-Bust Cycles

We also show how these loss functions can be easily modified to study the implications of the fed possibly falling behind shifts in the Phillips curve and generating a boom-bust cycle. More importantly, the loss-function approach provides a much better analytical tool for implementing state-of-the-art Forecasting and Policy Analysis Systems (FPAS) that focus on implementing a risk-management approach to monetary policy. The simple model of the U.S. economy is based on a paper by Alichi and others (2015) that recommended a modest overshooting strategy as part of the IMF’s Article IV US consultation.

Participants will be shown how to extend the model to include term structure equations and predictions for 1, 3, 5, and 10 year government bond yields.  They will also be shown how to create 2-country versions of the model to study exchange rate dynamics and nasty spillover effects from US monetary and fiscal policies to other countries.  In a nutshell, the depreciation in the US dollar caused by the US overshooting policy can result in an unwelcome appreciation in the currency for other countries with economic slack and inflation below target.

Crash Course on Divide-and-Conquer Strategies

Participants will also be given a crash course in understanding divide-and-conquer strategies to solve mixed-complementarity problems that involve nasty occasionally-binding constraints. Examples of occasionally-binding constraints include the Effective Lower Bound (ELB) on the Policy Rate as well as potentially efficient unconventional policy strategies that are designed to replace the policy space that is lost by hitting the ELB. This includes, for example, Yield Curve Caps (YCC) such as what has been deployed recently by the Reserve Bank of Australia and FX-intervention strategies (FXIS) as employed very successfully by the Czech National Bank to escape from a low inflation trap with significant risks of bad deflation. The former case involves communicating a time-varying restriction on the maximum level of market interest rates (at an appropriate duration!) while the latter involves a time-varying exchange rate floor. This course will be an important prerequisite for extensions to the basic model that will allow participants to study the potential benefits of strategies based on YCC and FXIS. It is also a prerequisite for a course on learning Bayesian estimation methods. 

Course will be of interest to …

Central-bank modelers interested in improving their existing Forecasting and Policy Analysis Systems or researchers and central-bank watchers interested in studying the risks of inflation scares. For documentation about the FPAS (Chapter 4 of "Advancing the Frontiers of Monetary Policy") and the 3 ingredients of a central bank forecast you can request a password access to our videos and training material (link to trial Page) by sending an email to douglaslaxton@thebetterpolicyproject.org.

In spite of the technical nature of the subject it will be taught on a level that is fairly accessible to a wide audience.

Warning! If you believe that it is only a question of designing a model and then hoping that the model will tell you the answers to questions, this course is definitely not for you. However, if you believe that models can be useful organizational devices to help impose macroeconomic consistency to study the implications of uncertainty then you might find this course and its extensions very useful. 


The course will last 4 days. Participants will have access to 4 hours of training each day. The courses are organized to focus on the needs of each central bank. So seats are very limited!

Don't worry! We will find a suitable time of the day for you. Participants will be divided into groups by their time zones.



  • DYNARE 4.63 or higher.

  • MATLAB 2020a or higher.


What will you learn?

  • Efficient Central Bank Forecasting and Policy Analysis Systems.

  • The 3 Ingredients of a Good Forecast.

  • Using DYNARE for both state-of-the-art practical monetary policy analysis and research designed to support the implementation of efficient central bank forecasting and policy analysis systems.

  • Solving Non-linear perfect foresight problems in DYNARE.

  • Solving Mixed Complementarity Problems in DYNARE.

  • Understanding robust and efficient divide-and-conquer algorithms for solving models with nasty occasionally binding constraints.

  • Brief introduction into using Python and R for data analysis and creating reports.

The outcomes from the course:

  • DYNARE code with closed-economy model of the U.S. Economy.

  • Monthly updates of U.S. medium-term scenarios produced by The Better Policy Project team.

  • Certificate of completion.

  • Further assistance (subject to discussion).

Reading Material







  • Laxton, Douglas, Hope Pioro, and Peter McAdam, 1996, “Solving MULTIMOD with First-Order and Newton-Based Techniques,” released as a University of Strathclyde Discussion Paper in Economic Modeling, October.


April 25
- April 28


Application Deadline:
April 18

June 13
- June 16


Application Deadline:
June 6

Registration Closed
Registration Open

Times will be discussed separately with the participants to adapt to their time-zones.

Price for 1 Participant

High-Income Countries -- €1000

Middle-Income Countries -- €500

Low-Income Countries -- €250

Countries are classified by income using new "World Bank country classifications by income level: 2020-2021." Find the World Bank tables here.