A major challenge in automating the production of a large number of forecasts, as often required in many business applications, is the need for robust and reliable predictions. Increased noise, outliers and structural changes in the series, all too common in practice, can severely affect the quality of forecasting. We investigate ways to increase the reliability of exponential smoothing forecasts, the most widely used family of forecasting models in business forecasting. We consider two alternative sets of approaches, one stemming from statistics and one from machine learning. To this end, we adapt Mestimators, boosting and inverse boosting to parameter estimation for exponential smoothing. We propose appropriate modifications that are necessary for time series forecasting while aiming to obtain scalable algorithms. We evaluate the various estimation methods using multiple real datasets and find that several approaches outperform the widely used maximum likelihood es-
This paper investigates the impact of the 2007 financial crisis on the relationship between real mortgage interest rates and real house prices. It applies a dynamic conditional correlation based methodology that uses fractionally differenced data along with controls for structural breaks and non-interest-rate related factors that influence house prices. The key finding made is that the financial crisis had a long-term structural impact on the monetary transmission relationship. For example, we find that the mean conditional correlation between house prices in England and Wales and the three-year fixed mortgage rate rose by 6.6 percentage points. Similarly, the mean correlation between prices and the standard variable mortgage rate increased 6.4 percentage points to 54%. These findings suggest to us that interest-rate-based monetary policy still has an important role to play in the housing market.
We investigate the nature of the relationship between Corporate Social Responsibility (CSR) and Corporate Financial Performance (CFP) by examining how it changes across a third dimension that accounts for firm-specific factors. We propose a semi-latent specification of an endogenous control variable, which can, for the first time, explicitly identify, for each individual firm, the threshold level where the marginal impact of CSR on CFP turns positive. We provide empirical evidence that this threshold depends on the additional dimension and consequently, the previously reported U-shape seems to be an aggregation of relationships of differential magnitude and direction. This disaggregation fits the data better and therefore, we maintain that the addition of a higher dimension, along with the identification of the threshold level, can explain the conflicting results in the literature.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.