Time Variation in Macro‐Financial Linkages

Abstract

Reproduction permitted only if source is stated.ISBN 978-3-86558-907-1 (Printversion) Non-technical summary None of the macroeconomic models commonly used in academic research and in policy institutions was able to predict the strong economic downturn following the Global Financial Crisis. Two main shortcomings of the standard macro modeling approach have been identi…ed: the lack of …nancial variables in these models and the lack of a time-varying relationship between …nancial and macroeconomic variables.We try to overcome these shortcomings and incorporate a few key …nancial indicators in an otherwise standard Bayesian macroeconomic vector autoregressive model (VAR) for the US and estimate that model over the period 1958Q1-2012Q2. The VAR includes GDP growth, GDP de ‡ator in ‡ation, house price in ‡ation, the corporate bond spread, stock price in ‡ation and the Federal Funds rate. In order to account for possible time variation in the relationship between …nancial indicators and the macroeconomy, we allow for continuous changes in the autoregressive coe¢ cients, contemporaneous relations and stochastic volatilities. This re ‡ects the fact that time variation in the shock transmission can occur because of permanent structural changes such as …nancial globalization, regulatory changes or changes in the conduct of monetary policy or because of temporary changes due to agency problems between lenders and borrowers which are more pronounced in …nancial crisis periods than in normal periods.Based on our estimated time-varying parameter VAR, we look at the sum of the contributions of shocks to each individual …nancial indicator to GDP growth as a measure of the overall importance of the …nancial sector as origin of shocks for the macroeconomy. We then shed light on the underlying sources of time variation. We assess the contribution of unexpected changes in individual …nancial variables to GDP growth over time and look at possible changes in the volatility of …nancial shocks and in their impact on GDP growth. Finally, we compare …nancial shock contributions estimated from the model with those estimated from a constant parameter VAR and a VAR in which we replace the …nancial variables with the National Financial Conditions Index published by the Federal Reserve Bank of Chicago, a latent factor extracted from a very large number of …nancial variables.Our main …ndings are: (i) The contribution of …nancial shocks to the forecast error variance of GDP growth ‡uctuates considerably over time, from about 20 percent in normal times to roughly 50 percent over the global …nancial crisis period. (ii) The Great Recession and the subsequent weak recovery can largely be traced back to negative housing shocks. (iii) Housing shocks have become more important for the real economy since the early-2000s, and negative housing shocks are more important than positive ones. Die wichtigsten Erkenntnisse lauten: a) Der Beitrag …nanzieller Schocks zur Prognosefehlervarianz des BIP-Wachstums schwankt im Zeitverlauf erheblich und reicht von etw...

Has the Fed Responded to House and Stock Prices? A Time-Varying Analysis

“…Prieto et al (2016) use data on several financial variables (including house prices and stock prices) to investigate the time-varying transmission mechanism and the relative importance of various financial shocks. However, they do not consider the systematic component of monetary policy in their analysis.…”

Credit constraints and the international propagation of US financial shocks

“…Our paper belongs to a growing group of empirical studies on macro-financial linkages ECB Working Paper 1954, August 2016(e.g., Lown and Morgan, 2006;Zakrajsek, 2011, 2012;Fornari and Stracca, 2012;Meeks, 2012;Prieto et al, 2016). Macro-financial models often fail to capture nonlinear amplification effects.…”

Financial Shocks and the Real Economy in a Nonlinear World: From Theory to Estimation

“…The two papers most closely related to our analysis are Ciccarelli, Ortega and Valderrama (2012) and Prieto, Eickmeier and Marcellino (2013), which are both based on TVP-VAR models. On the one hand, unlike Ciccarelli, Ortega and Valderrama (2012), we allow for stochastic volatility (whereas they assume time variation only in the autoregressive coefficient The variables in y t are the real GDP growth rate, the inflation rate, the short-term interest rate, the growth rate of the stock of credit to the private sector, and an indicator of financial stress especially designed for the euro area (the Composite Indicator of Systemic Stress, or CISS, based on the work of Holló, Kremer, and Lo Duca, 2012).…”