We propose a novel performance based forecast combination scheme. The scheme uses the standardized loss difference relative to the average model, thus accounting for the statistical magnitude of the losses. The theoretical goal of the scheme is to perform as well as the best candidate model, and we characterize the risk of the combination scheme relative to the best individual model.

Our scheme builds an intuitive bridge between model selection and forecast combination while showing robust performance in finite samples.

We apply the scheme to combine forecasts of financial tail risk, i.e., Value-at-Risk and expectiles. The data are large cap stocks from the NYSE Trade and Quote database. We find that the scheme is competitive, both with respect to alternative combination schemes and the individual models.

Real-time economic information is essential for policy-making but difficult to obtain. We introduce a granular nowcasting method for macro- and industry-level GDP using a network approach and data on real-time monthly inter-industry payments in the UK. To this purpose we devise a model which we call an extended generalised network autoregressive (GNAR-ex) model, tailored for networks with time-varying edge weights and nodal time series, that exploits the notion of neighbouring nodes and neighbouring edges. The performance of the model is illustrated on a range of synthetic data experiments. We implement the GNAR-ex model on the payments network including time series information of GDP and payment amounts. To obtain robustness against statistical revisions, we optimise the model over 9 quarterly releases of GDP data from the UK Office for National Statistics. Our GNAR-ex model can outperform baseline autoregressive benchmark models, leading to a reduced forecasting error. This work helps to obtain timely GDP estimates at the aggregate and industry level derived from alternative data sources compared to existing, mostly survey-based, methods. Thus, this paper contributes both, a novel model for networks with nodal time series and time-varying edge weights, and the first network-based approach for GDP nowcasting based on payments data.

This paper presents a novel copula-based no-arbitrage pricing framework for forecasting corporate bond returns and optimizing bond portfolios. Utilizing a copula-based dynamic factor model, we generate step-ahead forecasts for zero-coupon bond yields, which are subsequently applied to obtain and simulate the no-arbitrage prices for both callable and non-callable fixed-coupon bonds. These simulated bond prices serve as inputs for a novel convex multiobjective portfolio optimization, incorporating key criteria such as average returns, Conditional Value-at-Risk (CVaR), distance-to-default, transaction costs, and option-adjusted duration and convexity. Applying our methodology to a dataset of 879 corporate bonds denominated in Euros from January 2016 to July 2024, we demonstrate that the suggested copula-based no-arbitrage pricing framework takes advantage of the yield curve non-linear dependence structure and offers bond portfolios that consistently outperform those portfolios based on the classical dynamic Nelson-Siegel approach and an equally weighted (EQW) benchmark in terms of higher returns and Sharpe ratios while effectively reducing tail risk.

Index insurance design involves integrating weather data, soil moisture, phenology information, and satellite imagery, which presents challenges in data fusion. This article addresses the modeling of multisource functional indices of varying lengths stagewise by boosting an ensemble of sequential models. The implemented methods, including nonparametric regression and deep learning models, aim to improve crop yield prediction by accounting for spatiotemporal dependence. Results from an applied case study demonstrate the feasibility of stagewise modeling and the hedging effectiveness of the proposed index insurance contracts.