An Evaluation of Early Warning Models for Systemic Banking Crises: Does Machine Learning Improve Predictions?
Johannes Beutel, Sophia List, Gregor von Schweinitz
Abstract
This paper compares the out-of-sample predictive performance of different early warning models for systemic banking crises using a sample of advanced economies covering the past 45 years. We compare a benchmark logit approach to several machine learning approaches recently proposed in the literature. We find that while machine learning methods often attain a very high in-sample fit, they are outperformed by the logit approach in recursive out-of-sample evaluations. This result is robust to the choice of performance measure, crisis definition, preference parameter, and sample length, as well as to using different sets of variables and data transformations. Thus, our paper suggests that further enhancements to machine learning early warning models are needed before they are able to offer a substantial value-added for predicting systemic banking crises. Conventional logit models appear to use the available information already fairly effciently, and would for instance have been able to predict the 2007/2008 financial crisis out-of-sample for many countries. In line with economic intuition, these models identify credit expansions, asset price booms and external imbalances as key predictors of systemic banking crises.
Artikel Lesen
Does Machine Learning Help us Predict Banking Crises?
Johannes Beutel, Sophia List, Gregor von Schweinitz
Journal of Financial Stability,
December
2019
Abstract
This paper compares the out-of-sample predictive performance of different early warning models for systemic banking crises using a sample of advanced economies covering the past 45 years. We compare a benchmark logit approach to several machine learning approaches recently proposed in the literature. We find that while machine learning methods often attain a very high in-sample fit, they are outperformed by the logit approach in recursive out-of-sample evaluations. This result is robust to the choice of performance metric, crisis definition, preference parameter, and sample length, as well as to using different sets of variables and data transformations. Thus, our paper suggests that further enhancements to machine learning early warning models are needed before they are able to offer a substantial value-added for predicting systemic banking crises. Conventional logit models appear to use the available information already fairly efficiently, and would for instance have been able to predict the 2007/2008 financial crisis out-of-sample for many countries. In line with economic intuition, these models identify credit expansions, asset price booms and external imbalances as key predictors of systemic banking crises.
Artikel Lesen
Did the Swiss Exchange Rate Shock Shock the Market?
Manuel Buchholz, Gregor von Schweinitz, Lena Tonzer
Abstract
The Swiss National Bank abolished the exchange rate floor versus the Euro in January 2015. Based on a synthetic matching framework, we analyse the impact of this unexpected (and therefore exogenous) shock on the stock market. The results reveal a significant level shift (decline) in asset prices in Switzerland following the discontinuation of the minimum exchange rate. While adjustments in stock market returns were most pronounced directly after the news announcement, the variance was elevated for some weeks, indicating signs of increased uncertainty and potentially negative consequences for the real economy.
Artikel Lesen
The Macroeconomics of Testing and Quarantining
Martin S. Eichenbaum, Sergio Rebelo, Mathias Trabandt
Journal of Economic Dynamics and Control,
May
2022
Abstract
We develop a SIR-based macroeconomic model to study the impact of testing/ quarantining and social distancing/mask use on health and economic outcomes. These policies can dramatically reduce the costs of an epidemic. Absent testing/quarantining, the main effect of social distancing and mask use on health outcomes is to delay, rather than reduce, epidemic-related deaths. Social distancing and mask use reduce the severity of the epidemic-related recession but prolong its duration. There is an important synergy between social distancing and mask use and testing/quarantining. Social distancing and mask use buy time for testing and quarantining to come to the rescue. The benefits of testing/quarantining are even larger when people can get reinfected, either because the virus mutates or immunity is temporary.
Artikel Lesen
Involuntary Unemployment and the Business Cycle
Lawrence J. Christiano, Mathias Trabandt, Karl Walentin
Review of Economic Dynamics,
January
2021
Abstract
Can a model with limited labor market insurance explain standard macro and labor market data jointly? We construct a monetary model in which: i) the unemployed are worse off than the employed, i.e. unemployment is involuntary and ii) the labor force participation rate varies with the business cycle. To illustrate key features of our model, we start with the simplest possible framework. We then integrate the model into a medium-sized DSGE model and show that the resulting model does as well as existing models at accounting for the response of standard macroeconomic variables to monetary policy shocks and two technology shocks. In addition, the model does well at accounting for the response of the labor force and unemployment rate to these three shocks.
Artikel Lesen
Why They Keep Missing: An Empirical Investigation of Sovereign Bond Ratings and Their Timing
Gregor von Schweinitz, Makram El-Shagi
Scottish Journal of Political Economy,
Nr. 2,
2022
Abstract
Two contradictory strands of the rating literature criticize that rating agencies merely follow the market on the one hand, and emphasizing that rating changes affect capital movements on the other hand. Both focus on explaining rating levels rather than the timing of rating announcements. Contrarily, we explicitly differentiate between a decision to assess a country and the actual rating decision. We show that this differentiation significantly improves the estimation of the rating function. The three major rating agencies treat economic fundamentals similarly, while differing in their response to other factors such as strategic considerations. This reconciles the conflicting literature.
Artikel Lesen
Mission, Motivation, and the Active Decision to Work for a Social Cause
Sabrina Jeworrek, Vanessa Mertins
Nonprofit and Voluntary Sector Quarterly,
Nr. 2,
2022
Abstract
The mission of a job affects the type of worker attracted to an organization but may also provide incentives to an existing workforce. We conducted a natural field experiment with 246 short-term workers. We randomly allocated some of these workers to either a prosocial or a commercial job. Our data suggest that the mission of a job has a performance-enhancing motivational impact on particular individuals only, those with a prosocial attitude. However, the mission is very important if it has been actively selected. Those workers who have chosen to contribute to a social cause outperform the ones randomly assigned to the same job by about half a standard deviation. This effect seems to be a universal phenomenon that is not driven by information about the alternative job, the choice itself, or a particular subgroup.
Artikel Lesen
Offshoring, Domestic Employment and Production. Evidence from the German International Sourcing Survey
Wolfhard Kaus, Markus Zimmermann
IWH Discussion Papers,
Nr. 14,
2022
Abstract
This paper analyses the effect of offshoring (i.e., the relocation of activities previously performed in-house to foreign countries) on various firm outcomes (domestic employment, production, and productivity). It uses data from the International Sourcing Survey (ISS) 2017 for Germany, linked to other firm level data such as business register and ITGS data. First, we find that offshoring is a rare event: In the sample of firms with 50 or more persons employed, only about 3% of manufacturing firms and 1% of business service firms have performed offshoring in the period 2014-2016. Second, difference-in-differences propensity score matching estimates reveal a negative effect of offshoring on domestic employment and production. Most of this negative effect is not because the offshoring firms shrink, but rather because they don’t grow as fast as the non-offshoring firms. We further decompose the underlying employment dynamics by using direct survey evidence on how many jobs the firms destroyed/created due to offshoring. Moreover, we do not find an effect on labour productivity, since the negative effect on domestic employment and production are more or less of the same size. Third, the German data confirm previous findings for Denmark that offshoring is associated with an increase in the share of ‘produced goods imports’, i.e. offshoring firms increase their imports for the same goods they continue to produce domestically. In contrast, it is not the case that offshoring firms increase the share of intermediate goods imports (a commonly used proxy for offshoring), as defined by the BEC Rev. 5 classification.
Artikel Lesen
Sovereign Default Risk, Macroeconomic Fluctuations and Monetary-Fiscal Stabilisation
Markus Kirchner, Malte Rieth
IWH Discussion Papers,
Nr. 22,
2020
Abstract
This paper examines the role of sovereign default beliefs for macroeconomic fluctuations and stabilisation policy in a small open economy where fiscal solvency is a critical problem. We set up and estimate a DSGE model on Turkish data and show that accounting for sovereign risk significantly improves the fit of the model through an endogenous amplication between default beliefs, exchange rate and inflation movements. We then use the estimated model to study the implications of sovereign risk for stability, fiscal and monetary policy, and their interaction. We find that a relatively strong fiscal feedback from deficits to taxes, some exchange rate targeting, or a monetary response to default premia are more effective and efficient stabilisation tools than hawkish inflation targeting.
Artikel Lesen
Disentangling Covid-19, Economic Mobility, and Containment Policy Shocks
Annika Camehl, Malte Rieth
IWH Discussion Papers,
Nr. 2,
2021
Abstract
We study the dynamic impact of Covid-19, economic mobility, and containment policy shocks. We use Bayesian panel structural vector autoregressions with daily data for 44 countries, identified through sign and zero restrictions. Incidence and mobility shocks raise cases and deaths significantly for two months. Restrictive policy shocks lower mobility immediately, cases after one week, and deaths after three weeks. Non-pharmaceutical interventions explain half of the variation in mobility, cases, and deaths worldwide. These flattened the pandemic curve, while deepening the global mobility recession. The policy tradeoff is 1 p.p. less mobility per day for 9% fewer deaths after two months.
Artikel Lesen
