The Time-Varying Risk of Macroeconomic Disasters

Abstract

While time-varying disasters can explain many characteristics of financial markets, their quantitative assessment is still missing. We propose a latent variable approach to estimate the time-varying probability of a macroeconomic disaster, using a dataset of 42 countries over more than 100 years. We find that disaster risk is volatile and persistent, strongly correlates with the dividend yield, and forecasts stock returns. A state-of-the-art model calibrated with our disaster risk estimates generates a large and volatile equity premium and a low risk free rate under standard assumptions. This evidence supports the idea that investors' fear of disasters drives equity premium dynamics.

Neural networks for estimating Macro Asset Pricing model in football clubs

“…However, these results confirm the superiority of QNN and DNDT, which improve precision results compared to other previous studies on estimating Asset Pricing models (Chen et al, 2017; Gu et al, 2020, 2021; Marfè & Penasse, 2016). For example, Marfè and Penasse (2016) obtained an 86% ROC curve to exclusively predict disaster risk in the price of shares. Chen et al (2017) obtained a mean goodness of fit in their different estimated parameters of 30% using the OLS technique for their estimation.…”

“…The specification used for the volatility, consumption, and dividend parameters are equivalent to those applied in previous works (Chen et al, 2017; Marfè & Penasse, 2016; Schorfheide et al, 2018). Typically, volatility is found as a variable with a positive sign (Gu et al, 2021).…”

“…Asset pricing models have been developed mostly with conventional statistical linear methods, mainly using linear regression. There are few experiences of the application of non‐linear methods in the estimation of these models (Gu et al, 2021), although they have shown superior precision results and have been limited to US stock indices, some new experiences that show the capacity of success that these non‐linear methods have on asset pricing (Brunnermeier et al, 2021; Marfè & Penasse, 2016; Särkkä, 2013; Schorfheide et al, 2018). Other authors have researched the football industry but have applied other methodologies compared to the use of machine learning techniques in our study.…”

Changes in Natural Disaster Risk: Macroeconomic Responses in Selected Latin American Countries

“…Isoré and Szczerbowicz (2017) further showed how to extend this approach to a New Keynesian environment, and found that disaster risk shocksn-again, absent of actual disaster realization-, generate procyclical responses of consumption, investment, labor, wage, and inflation, simultaneously to the recession and rise in equity premium. Further empirical evidence by Siriwardane (2015) and Marfè and Penasse (2017) support the relationship between changes in the probability of a disaster and macroeconomic variables.…”

Market Fragility and the Paradox of the Recent Stock-Bond Dissonance

“…More speci…cally we refer to Gabaix (2012), Gourio (2012), andWachter (2013), who demonstrate that this variability can explain many asset-pricing puzzles. In addition, Marfe and Penasse (2017) …nd empirical evidence for disaster-risk variablity. Another body of literature assumes imperfect information about rare disaster risk and argues that parameter learning implies more pessimistic disaster-risk beliefs after a rare disaster (Collin-Dufresne et al, 2016, Koulovatianos and Wieland, 2017, and Kozlowski et al, 2017.…”