International Journal of Forecasting最新文献

Forecast combinations have flourished remarkably in the forecasting community and, in recent years, have become part of mainstream forecasting research and activities. Combining multiple forecasts produced for a target time series is now widely used to improve accuracy through the integration of information gleaned from different sources, thereby avoiding the need to identify a single “best” forecast. Combination schemes have evolved from simple combination methods without estimation to sophisticated techniques involving time-varying weights, nonlinear combinations, correlations among components, and cross-learning. They include combining point forecasts and combining probabilistic forecasts. This paper provides an up-to-date review of the extensive literature on forecast combinations and a reference to available open-source software implementations. We discuss the potential and limitations of various methods and highlight how these ideas have developed over time. Some crucial issues concerning the utility of forecast combinations are also surveyed. Finally, we conclude with current research gaps and potential insights for future research.

We provide a comprehensive assessment of the predictive power of combinations of dynamic stochastic general equilibrium (DSGE) models for GDP growth, inflation, and the interest rate in the euro area. We employ a battery of static and dynamic pooling weights based on Bayesian model averaging principles, prediction pools, and dynamic factor representations, and entertain six different DSGE specifications and five prediction weighting schemes. Our results indicate that exploiting mixtures of DSGE models produces competitive forecasts compared to individual specifications for both point and density forecasts over the last three decades. Although these combinations do not tend to systematically achieve superior forecast performance, we find improvements for particular periods of time and variables when using prediction pooling, dynamic model averaging, and combinations of forecasts based on Bayesian predictive synthesis.

Any lead–lag effect in an asset pair implies that future returns on the lagging asset have the potential to be predicted from past and present prices of the leader, thus creating statistical arbitrage opportunities. We utilize robust lead–lag indicators to uncover the origin of price discovery, and we propose an econometric model exploiting that effect with level 1 data of limit order books (LOBs). We also develop a high-frequency trading strategy based on the model predictions to capture arbitrage opportunities. The framework is then evaluated on six months of DAX 30 cross-listed stocks’ LOB data obtained from three European exchanges in 2013: Xetra, Chi-X, and BATS. We show that a high-frequency trader can profit from lead–lag relationships because of predictability, even when trading costs, latency, and execution-related risks are considered.

Uncertainty variables involving diverse aspects play leading roles in determining oil price movements. This study aims to improve the aggregate crude oil market volatility prediction based on a large set of uncertainty variables from a comprehensive viewpoint. Specifically, we apply three shrinkage methods, namely, forecast combination, dimension reduction, and variable selection, to extract valuable predictive information in a data-rich world. The empirical results show that the forecasting power of the individual uncertainty index is not satisfactory. By contrast, all shrinkage models, particularly the supervised machine learning techniques, demonstrate outstanding predictability of oil market volatility, which tends to be strong during business recessions. Notably, the sizeable economic gains confirm the superior forecasting performance of our comprehensive framework. We provide solid evidence that the two option-implied volatility variables uniformly serve as the best two predictors.

Countries experiencing balance of payments (BOP) crises may obtain IMF loans to stabilize external accounts. These loans require IMF programs that outline performance targets to ensure forecasted recovery trajectories. Two key indicators of external account performance are reserves and short-term external debt (“STdebt”). Extensive literature evaluates IMF forecasts, but reserves and STdebt have not been studied. We construct a database of nearly 300 BOP crisis countries with IMF BOP programs from 1992–2019. Reserve forecasts are shown to be systematically biased and inefficient, a result that is startlingly persistent across (a) degrees of capital mobility, (b) trade openness, (c) exchange rate regimes, (d) inflation, and (e) country income levels. We show the bias is driven by deeply pessimistic IMF reserve forecasts that underestimate reserves and systematically ignore information known at the time of the forecast. STdebt forecasts are also inefficient but with an optimistic bias, systematically underestimating future debt. If STdebt is used to peg reserve requirements, the optimistic bias of STdebt forecasts may drive the pessimistic bias of reserve forecasts.

We develop new methods for testing equal predictive accuracy for panels of forecasts, exploiting information in both the time-series and cross-sectional dimensions of the data. We examine general tests of equal forecasting performance averaged across all time periods and individual units, along with tests that focus on subsets of time or clusters of units. Properties of our tests are demonstrated through Monte Carlo simulations and in an empirical application that compares International Monetary Fund forecasts of country-level real gross domestic product growth and inflation to private-sector survey forecasts and forecasts from a simple time-series model.

The ability to accurately predict future events is critical in numerous areas of human life. Past research has shown that human reasoning can usefully predict geopolitical outcomes, but such forecasts are still far from perfect. In the current work, we investigate whether machine learning can help predict whether people’s forecasts are likely to be correct. We rely on data from a geopolitical forecasting contest where participants provided a total of 1530 predictions accompanied by written rationales. We extracted various features (e.g., forecasters’ psychological traits, the linguistic aspects of the rationales, and peer evaluations), trained a machine learning model to predict the accuracy of prediction, and validated it on held-out data. The results showed that the model was able to predict the accuracy of a prediction with excellent accuracy. A theoretical simulation shows that aggregating predictions based on the output of our prediction model can yield highly accurate forecasts. We conclude that combining human intelligence with machine learning algorithms can make the future more predictable.

Mixed-frequency data prediction tasks are pertinent in various application domains, in which one leverages progressively available high-frequency data to forecast/nowcast the low-frequency ones. Existing methods in the literature tailored to such tasks are mostly linear in nature; depending on the specific formulation, they largely rely on the assumption that the (latent) processes that govern the dynamics of the high- and low-frequency blocks of variables evolve at the same frequency, either the low or the high one. This paper develops a neural network-based multi-task shared-encoder-dual-decoder framework for joint multi-horizon prediction of both the low- and high-frequency blocks of variables, wherein the encoder/decoder modules can be either long short-term memory or transformer ones. It addresses forecast/nowcast tasks in a unified manner, leveraging the encoder–decoder structure that can naturally accommodate the mixed-frequency nature of the data. The proposed framework exhibited competitive performance when assessed on both synthetic data experiments and two real datasets of US macroeconomic indicators and electricity data.

We propose a novel method for forecasting network time series that occur in graphs or networks. Our approach is based on a spectral graph wavelet transform (SGWT) that provides the localized behavior of graph signals around each node. The proposed method improves forecasting performance over other existing methods. In particular, the advantages of the proposed method stand out when signals observed on a graph are inhomogeneous or non-stationary. We demonstrate the strength of the proposed approach through real-world data analysis. This analysis uses two network time series datasets: the daily number of people getting on and off the Seoul Metropolitan Subway, and daily Covid-19 confirmed cases reported in South Korea. We further conduct a simulation study to evaluate the effectiveness of the proposed method.

In this paper, we assess whether and when multi-country studies pay off for forecasting inflation and output growth. Factor stochastic volatility is adopted to allow for cross-country linkages and model economies jointly. We estimate factors and rely on post-processing, rather than expert judgement, to obtain an estimate for the number of factors. This is different from most existing two-step approaches in the factor literature. Our approach is then used to extend the existing unobserved components model, which assumes that 34 economies are independent. The results suggest that allowing for cross-country linkages yields inflation and output growth forecasts that are highly competitive with those obtained from estimating economies independently. Zooming into the forecast performance over time reveals that allowing for cross-country linkages is of particular importance when interest centres on forecasting periods of uncertainty. Another key finding is that the estimated global factors are correlated with the domestic business cycle. We interpret this to mean that part of the variation captured in global factors reflects a global business cycle.