« Back to Results

Macro Modelling, Financial Crisis and Monetary Policy

Paper Session

Sunday, Jan. 3, 2021 3:45 PM - 5:45 PM (EST)

Hosted By: Society for Nonlinear Dynamics and Econometrics
  • Chair: Hilde C. Bjørnland, BI Norwegian Business School

The Macroeconomy as a Random Forest

Philippe Goulet Coulombe
,
University of Pennsylvania

Abstract

I develop Macroeconomic Random Forest (MRF), an algorithm adapting the canonical Machine Learning (ML) tool to flexibly model evolving parameters in a linear macro equation. Its main output, Generalized Time-Varying Parameters (GTVPs), is a versatile device nesting many popular nonlinearities (threshold/switching, smooth transition, structural breaks/change) and allowing for sophisticated new ones. The approach delivers clear forecasting gains over numerous alternatives, predicts the 2008 drastic rise in unemployment, and performs well for inflation. Unlike most ML-based methods, MRF is directly interpretable — via its GTVPs. For instance, the successful unemployment forecast is due to the influence of forward-looking variables (e.g., term spreads, housing starts) nearly doubling before every recession. Interestingly, the Phillips curve has indeed flattened, and its might is highly cyclical.

Noisy Monetary Policy

Tatjana Dahlhaus
,
Bank of Canada
Luca Gambetti
,
University of Turin

Abstract

We introduce imperfect information in monetary policy (in the sense of imperfect central bank communications). Agents receive signals from the central bank revealing new information ("news") about the future evolution of the policy rate before changes in the rate actually take place. However, the signal is disturbed by noise. We employ a non-standard vector autoregression procedure to disentangle the economic and financial effects of news and noise in US monetary policy since the mid-1990s. Using survey- and market-based data on federal funds rate expectations, we find that the noisy signal plays a relatively important role for macroeconomic dynamics. A signal reporting news about a future policy tightening shifts policy rate expectations upwards and decreases output and prices. A sizable part of the signal is noise surrounding future monetary policy actions. The noise decreases output and prices and can explain up to 16% and 13% of their variations, respectively. Furthermore, it significantly increases the excess bond premium, the corporate spread and financial market volatility, and decreases stock prices.

Estimating Macroeconomic Models of Financial Crises: An Endogenous Regime-Switching Approach

Christopher Otrok
,
University of Missouri
Gianluca Benigno
,
London School of Economics
Andrew Foerster
,
Federal Reserve Bank of San Francisco
Alessandro Rebucci
,
Johns Hopkins University

Abstract

We estimate a workhorse DSGE model with an occasionally binding borrowing constraint. First, we propose a new specification of the occasionally binding constraint, where the transition between the unconstrained and constrained states is a stochastic function of the leverage level and the constraint multiplier. This specification maps into an endogenous regime-switching model. Second, we develop a general perturbation method for the solution of such a model. Third, we estimate the model with Bayesian methods to fit Mexico's business cycle and financial crisis history since 1981. The estimated model fits the data well, identifying three crisis episodes of varying duration and intensity: the Debt Crisis in the early-1980s, the Peso Crisis in the mid-1990s, and the Global Financial Crisis in the late-2000s. The crisis episodes generated by the estimated model display sluggish and long-lasting build-up and stagnation phases driven by plausible combinations of shocks. Different sets of shocks explain different variables over the
business cycle and the three historical episodes of sudden stops identified.

The Interaction Between Credit Constraints and Uncertainty Shocks

Pratiti Chatterjee
,
University of New South Wales
David Gunawan
,
University of Wollongong
Robert Kohn
,
University of New South Wales

Abstract

We propose a novel link between credit markets and uncertainty shocks. Empirically, we estimate time-varying uncertainty about credit in the U.S. and decompose it into a “pure” (independent) second-moment shock versus a second-moment change that is correlated with a first-moment shock. We show that a pure second-moment shock has almost no effect in expansions, but generates a significant decline across measures of real activity in recessions. To build intuition, we feed our estimated uncertainty process into a flexible-price real business cycle model with collateral constraints. A shock to credit uncertainty triggers a precautionary response that interacts with the collateral constraint to generate a sizable and simultaneous decline in output, consumption, investment, real wages, and hours. Our analysis thus provides a direct link between the source and manifestation of uncertainty leading to a better understanding about the transmission of uncertainty to the real economy.
Discussant(s)
Benjamin Wong
,
Monash University
Leif Anders Thorsrud
,
BI Norwegian Business School
Yoosoon Chang
,
Indiana University
Efrem Castelnuovo
,
University of Padova
JEL Classifications
  • E3 - Prices, Business Fluctuations, and Cycles
  • C3 - Multiple or Simultaneous Equation Models; Multiple Variables