New Keynesian DSGE Models, Financial Frictions and Bayesian Estimation


Lawrence J. Christiano



I plan to review the basic New Keynesian model and an extension that takes into account financial frictions.  The course is aimed at a broad audience, including people actively doing research with dynamic, stochastic, general equilibrium (DSGE) models, as well as people interested in seeing a review of the structure of these models and what they are used for. There will be afternoon practicum sessions. The sessions are designed to acquaint participants with Dynare as a tool for analyzing, solving and estimating DSGE models. The first part of these sessions is integrally related to the lectures (especially (1) below), as they explore the fundamental properties and policy implications of the New Keynesian model. In the second part of the afternoon sessions, we will review the fundamentals of Bayesian inference and then do Bayesian inference using Dynare.


Three Morning Lectures


1)    The simple New Keynesian (NK) model without capital (longer version of slides; background: my handbook chapter, and this comment on Acemoglu-Akcigit-Kerr). Will stress the impact of networks on the cost of inflation, the slope of the Phillips curve and the value of the Taylor Principle for inflation stabilization. Dynare code for the model is here.

a)    The Best (‘Natural’) equilibrium, Ramsey Equilibrium.

b)    The linearized Phillips curve and networks.

c)     Solving the model by linearization.


2)     Introducing financial frictions into the New Keynesian DSGE Model.

a)    Microfoundations for the Costly State Verification (CSV) approach (longer handout; section 6 of background paper).

b)    Integrating CSV into an NK model and the results of Bayesian estimation of the model using US and EA data (manuscript).

i)       The model.

ii)    The importance of risk shocks.

iii)  The response of monetary policy to an increase in interest rate spreads.

iv)  Background reading: Bernanke, Gertler and Gilchrist’s classic 1999 paper and Christiano-Motto-Rostagno.

c)     Extension to small open economy (manuscript and code, slides).

d)    Extension by Copaciu-Nalban-Bulete, addressing the interaction of possible currency mismatch problems in emerging markets that are anticipated as the US Federal Reserve implements ‘lift off’.

e)    Additional material on financial frictions specifically in the banking sector (slides, reading).


3)    Lecture  on Bayesian inference.


Three Afternoon Sessions

 In the afternoon sessions, participants can work with Dynare programs to explore: (i) basic economic principles implied by the New Keynesian model and (ii) methods for the empirical analysis of DSGE models, including Bayesian inference. Part (i) will build on part 1) of the morning lectures. Part (ii) will be preceded by the lecture on Bayesian inference. The afternoon sessions will center on the questions in the tutorial.

Apart from giving participants hands-on experience with the quantitative analysis of models using Dynare, question 1 in part 2 of the tutorial allows us to discuss the following topics using the model developed in the first lecture: 

1)    The sensitivity of the dynamic response of inflation and output to the persistence properties of shocks.

a)    Making precise the NK concepts of ‘insufficient aggregate demand’ and ‘excessive aggregate demand’ (see section 3.4 of handbook chapter).

2)    The Taylor principle.

a)    The rationale for the principle in the standard NK model (see section 3.1 of handbook chapter).

b)    The Taylor rule moves the interest rate in the right direction in response to ‘standard’ shocks, but does not move it far enough (see section 3.4 of handbook chapter).

3)    Circumstances when things can go awry with the Taylor principle:

a)    An important working capital channel may overturn the stabilizing properties of the Taylor principle (section 3.1 of handbook chapter).

b)    News shocks may imply that the monetary authority implementing the Taylor principle moves the interest rate in the wrong direction (see the following slides; Christiano-Ilut-Motto-Rostagno, Jackson Hole paper; and section 3.2 of handbook chapter).