报告一题目：Time-varying Model Averaging
报告摘要：Structural changes often occur in economics and finance due to changes in preferences,technologies, institutional arrangements, policies, crises, etc. Improving forecast accuracy ofeconomic time series withstructural changes is a long-standing problem. Model averagingaims at providing an insurance against selecting a poor forecast model. Allexisting modelaveraging approaches in the literature are designed with constant (non-time-varying)combination weights. Little attention has been paid to time-varying model averaging, whichis more realistic in economics under structural changes. This paper proposes a novel modelaveraging estimator which selects optimal time-varying combination weights by minimizinga local jackknife criterion. It is shown that the proposed time-varying jackknife modelaveraging (TVJMA) estimator is asymptotically optimal in the sense of achieving the lowestpossible local squared error loss in a class of time-varying model averaging estimators. Under a set of regularity assumptions, the TVJMA estimator is root-Th consistent. A simulationstudy and an empirical application highlight the merits of the proposed TVJMA estimatorrelative to a variety of popular estimators with constant model averaging weights and modelselection.
报告二题目：A two-step method for interval-valued crude oil price forecasting
报告摘要：As crude oil price is influenced by numerous factors, capturing itsbehavior precisely isquite challenging, and thus leads to the difficulty of forecasting. The current studies ignore the extreme nature of the lower and upper bounds of crude oil price. We propose atwo-step forecasting procedure that uses two techniques to consider the relevant informationavailable in the interval format. First, we extend the L 2 Boosting byBühlmann (2006) tothe interval-valued data to achieve variable selection for high-dimensional data. Second, aleave-subject-out cross-validation model averaging (LsoMA) method by Liao et al. (2019)is extended to average predictions from interval models with interval-valued exogenous variables. The empirical results show that our proposed approachsignificantly outperforms thebenchmark models in terms of both forecasting accuracy and robustness analysis.