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
Exchange Rate Regime, Real Misalignment and Currency Crises
Oliver Holtemöller, Sushanta Mallick
Economic Modelling,
Nr. 34,
2013
Abstract
Based on 69 sample countries, this paper examines the effect of macroeconomic fundamentals on real effective exchange rates (REER) in these sample countries. Using the misalignment of actual REER from its equilibrium level, we have estimated the factors explaining the extent of currency over- or under-valuation. Overall, we find that the higher the flexibility of the currency regime, the lower is the misalignment. The estimates are robust to different sub-samples of countries. We then explore the impact of such misalignment on the probability of a currency crisis in the next period, indicating the extent to which misalignment could be used as a leading indicator of a potential crisis. This paper thus makes a new contribution to the debate on the choice of exchange rate regime by bringing together real exchange rate misalignment and currency crisis literature.
Artikel Lesen
The Transition to Post-industrial BMI Values among US Children
Marco Sunder, Ariane Breitfelder, John Komlos
American Journal of Human Biology,
2009
Abstract
The trend in the BMI values of US children has not been estimated very convincingly because of the absence of longitudinal data. Our objective is to estimate time series of BMI values by birth cohorts instead of measurement years. We use five regression models to estimate the BMI trends of non-Hispanic US-born black and white children and adolescents ages 2-19 between 1941 and 2004. The increase in BMIZ values during the period considered was 1.3 (95% CI: 1.16; 1.44) among black girls, 0.8 for black boys, 0.7 for white boys, and 0.6 for white girls. This translates into an increase in BMI values of some 5.6, 3.3, 2.4, and 1.5 units, respectively. While the increase in BMI values started among the birth cohorts of the 1940s among black girls, the rate of increase tended to accelerate among all four ethnic/gender groups born in the mid-1950s to early-1960s. Some regional evidence leads to the conjecture that the spread of automobiles and radios affected the BMI values of boys already in the interwar period. We suppose that the changes in lifestyle associated with the labor saving technological developments of the 20th century are associated with the weight gains observed. The increased popularity of television viewing was most prominently associated with the contemporaneous acceleration in BMI gain. Am. J. Hum. Biol., 2009. © 2008 Wiley-Liss, Inc.
Artikel Lesen
