Forecasting Economic Activity Using a Neural Network in Uncertain Times: Monte Carlo Evidence and Application to the
German GDP
Oliver Holtemöller, Boris Kozyrev
IWH Discussion Papers,
Nr. 6,
2024
Abstract
In this study, we analyzed the forecasting and nowcasting performance of a generalized regression neural network (GRNN). We provide evidence from Monte Carlo simulations for the relative forecast performance of GRNN depending on the data-generating process. We show that GRNN outperforms an autoregressive benchmark model in many practically relevant cases. Then, we applied GRNN to forecast quarterly German GDP growth by extending univariate GRNN to multivariate and mixed-frequency settings. We could distinguish between “normal” times and situations where the time-series behavior is very different from “normal” times such as during the COVID-19 recession and recovery. GRNN was superior in terms of root mean forecast errors compared to an autoregressive model and to more sophisticated approaches such as dynamic factor models if applied appropriately.
Artikel Lesen
Does IFRS Information on Tax Loss Carryforwards and Negative Performance Improve Predictions of Earnings and Cash Flows?
Sandra Dreher, Sebastian Eichfelder, Felix Noth
Journal of Business Economics,
January
2024
Abstract
We analyze the usefulness of accounting information on tax loss carryforwards and negative performance to predict earnings and cash flows. We use hand-collected information on tax loss carryforwards and corresponding deferred taxes from the International Financial Reporting Standards tax footnotes for listed firms from Germany. Our out-of-sample tests show that considering accounting information on tax loss carryforwards does not enhance performance forecasts and typically even worsens predictions. The most likely explanation is model overfitting. Besides, common forecasting approaches that deal with negative performance are prone to prediction errors. We provide a simple empirical specification to account for that problem.
Artikel Lesen
Conditional Macroeconomic Survey Forecasts: Revisions and Errors
Alexander Glas, Katja Heinisch
Journal of International Money and Finance,
November
2023
Abstract
Using data from the European Central Bank's Survey of Professional Forecasters and ECB/Eurosystem staff projections, we analyze the role of ex-ante conditioning variables for macroeconomic forecasts. In particular, we test to which extent the updating and ex-post performance of predictions for inflation, real GDP growth and unemployment are related to beliefs about future oil prices, exchange rates, interest rates and wage growth. While oil price and exchange rate predictions are updated more frequently than macroeconomic forecasts, the opposite is true for interest rate and wage growth expectations. Beliefs about future inflation are closely associated with oil price expectations, whereas expected interest rates are related to predictions of output growth and unemployment. Exchange rate predictions also matter for macroeconomic forecasts, albeit less so than the other variables. With regard to forecast errors, wage growth and GDP growth closely comove, but only during the period when interest rates are at the effective zero lower bound.
Artikel Lesen
Department Profiles
Research Profiles of the IWH Departments All doctoral students are allocated to one of the four research departments (Financial Markets – Laws, Regulations and Factor Markets –…
Zur Seite
Projekte
Unsere Projekte 07.2022 ‐ 12.2026 Evaluierung des InvKG und des Bundesprogrammes STARK Bundesministerium für Wirtschaft und Klimaschutz (BMWK) Im Auftrag des Bundesministeriums…
Zur Seite
Department Profiles
Research Profiles of the IWH Departments All doctoral students are allocated to one of the four research departments (Financial Markets – Laws, Regulations and Factor Markets –…
Zur Seite
Conditional Macroeconomic Forecasts: Disagreement, Revisions and Forecast Errors
Alexander Glas, Katja Heinisch
IWH Discussion Papers,
Nr. 7,
2021
Abstract
Using data from the European Central Bank‘s Survey of Professional Forecasters, we analyse the role of ex-ante conditioning variables for macroeconomic forecasts. In particular, we test to which extent the heterogeneity, updating and ex-post performance of predictions for inflation, real GDP growth and the unemployment rate are related to assumptions about future oil prices, exchange rates, interest rates and wage growth. Our findings indicate that inflation forecasts are closely associated with oil price expectations, whereas expected interest rates are used primarily to predict output growth and unemployment. Expectations about exchange rates and wage growth also matter for macroeconomic forecasts, albeit less so than oil prices and interest rates. We show that survey participants can considerably improve forecast accuracy for macroeconomic outcomes by reducing prediction errors for external conditions. Our results contribute to a better understanding of the expectation formation process of experts.
Artikel Lesen
Optimizing Policymakers’ Loss Functions in Crisis Prediction: Before, Within or After?
Peter Sarlin, Gregor von Schweinitz
Macroeconomic Dynamics,
Nr. 1,
2021
Abstract
Recurring financial instabilities have led policymakers to rely on early-warning models to signal financial vulnerabilities. These models rely on ex-post optimization of signaling thresholds on crisis probabilities accounting for preferences between forecast errors, but come with the crucial drawback of unstable thresholds in recursive estimations. We propose two alternatives for threshold setting with similar or better out-of-sample performance: (i) including preferences in the estimation itself and (ii) setting thresholds ex-ante according to preferences only. Given probabilistic model output, it is intuitive that a decision rule is independent of the data or model specification, as thresholds on probabilities represent a willingness to issue a false alarm vis-à-vis missing a crisis. We provide real-world and simulation evidence that this simplification results in stable thresholds, while keeping or improving on out-of-sample performance. Our solution is not restricted to binary-choice models, but directly transferable to the signaling approach and all probabilistic early-warning models.
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
Should Forecasters Use Real‐time Data to Evaluate Leading Indicator Models for GDP Prediction? German Evidence
Katja Heinisch, Rolf Scheufele
German Economic Review,
Nr. 4,
2019
Abstract
In this paper, we investigate whether differences exist among forecasts using real‐time or latest‐available data to predict gross domestic product (GDP). We employ mixed‐frequency models and real‐time data to reassess the role of surveys and financial data relative to industrial production and orders in Germany. Although we find evidence that forecast characteristics based on real‐time and final data releases differ, we also observe minimal impacts on the relative forecasting performance of indicator models. However, when obtaining the optimal combination of soft and hard data, the use of final release data may understate the role of survey information.
Artikel Lesen