Value-added forecasting

Long ago, the emphasis shifted away from forecasting as a competitive weapon when it became apparent that forecast error could never be eliminated. Forecasts became a necessary evil that no one wanted to claim responsibility for. It's time to clear up some of the misconceptions about forecasts and to seize the opportunity inherent in the forecasting process. It is not forecast accuracy but rather improved understanding and use of forecasting as a tool for reducing both costs and lead times that will add real value to an enterprise and can improve the results from any and all other initiatives.     

Theory-coherent forecasting

We consider a method for producing multivariate density forecasts that satisfy moment restrictions implied by economic theory, such as Euler conditions. The method starts from a base forecast that might not satisfy the theoretical restrictions and forces it to satisfy the moment conditions using exponential tilting. Although exponential tilting has been considered before in a Bayesian context (Robertson et al. 2005), our main contributions are: (1) to adapt the method to a classical inferential context with out-of-sample evaluation objectives and parameter estimation uncertainty; and (2) to formally discuss the conditions under which the method delivers improvements in forecast accuracy. An empirical illustration which incorporates Euler conditions into forecasts produced by Bayesian vector autoregressions shows that the improvements in accuracy can be sizable and significant.     

Portfolio optimization based on GARCH-EVT-Copula forecasting models

This study uses GARCH-EVT-copula and ARMA-GARCH-EVT-copula models to perform out-of-sample forecasts and simulate one-day-ahead returns for ten stock indexes. We construct optimal portfolios based on the global minimum variance (GMV), minimum conditional value-at-risk (Min-CVaR) and certainty equivalence tangency (CET) criteria, and model the dependence structure between stock market returns by employing elliptical (Student-$t$ and Gaussian) and Archimedean (Clayton, Frank and Gumbel) copulas. We analyze the performances of 288 risk modeling portfolio strategies using out-of-sample back-testing. Our main finding is that the CET portfolio, based on ARMA-GARCH-EVT-copula forecasts, outperforms the benchmark portfolio based on historical returns. The regression analyses show that GARCH-EVT forecasting models, which use Gaussian or Student-$t$ copulas, are best at reducing the portfolio risk.     

Stochastic demographic forecasting

"This paper describes a particular approach to stochastic population forecasting, which is implemented for the U.S.A. through 2065. Statistical time series methods are combined with demographic models to produce plausible long run forecasts of vital rates, with probability distributions. The resulting mortality forecasts imply gains in future life expectancy that are roughly twice as large as those forecast by the Office of the Social Security Actuary....Resulting stochastic forecasts of the elderly population, elderly dependency ratios, and payroll tax rates for health, education and pensions are presented."     

Influence of differentiated roles on group forecasting accuracy

While behavioral research on forecasting has mostly examined the individual forecaster, organizationally-based forecasting processes typically tend to rely on groups with members from different functional areas for arriving at ‘consensus’ forecasts. The forecasting performance could also vary depending on the particular group structuring utilized in reaching a final prediction. The current study compares the forecasting performance of modified consensus groups with that of staticized groups using formal role-playing. It is found that, when undistorted model forecasts are given, group forecasts (whether they are arrived at through averaging or by a detailed discussion of the forecasts) contribute positively to the forecasting accuracy. However, providing distorted initial forecasts affects the final accuracy with varying degrees of improvement over the initial forecasts. The results show a strong tendency to favor optimistic forecasts for both the staticized and modified consensus group forecasts. Overall, the role modifications are found to be successful in eliciting a differential adjustment behavior, effectively mimicking the disparities between different organizational roles. Current research suggests that group discussions may be an efficient method of displaying and resolving differential motivational contingencies, potentially leading to group forecasts that perform quite well.     

Short-term forecasting of crime

The major question investigated is whether it is possible to accurately forecast selected crimes 1 month ahead in small areas, such as police precincts. In a case study of Pittsburgh, PA, we contrast the forecast accuracy of univariate time series models with naïve methods commonly used by police. A major result, expected for the small-scale data of this problem, is that average crime count by precinct is the major determinant of forecast accuracy. A fixed-effects regression model of absolute percent forecast error shows that such counts need to be on the order of 30 or more to achieve accuracy of 20% absolute forecast error or less. A second major result is that practically any model-based forecasting approach is vastly more accurate than current police practices. Holt exponential smoothing with monthly seasonality estimated using city-wide data is the most accurate forecast model for precinct-level crime series.     

The tourism forecasting competition

We evaluate the performances of various methods for forecasting tourism data. The data used include 366 monthly series, 427 quarterly series and 518 annual series, all supplied to us by either tourism bodies or academics who had used them in previous tourism forecasting studies. The forecasting methods implemented in the competition are univariate and multivariate time series approaches, and econometric models. This forecasting competition differs from previous competitions in several ways: (i) we concentrate on tourism data only; (ii) we include approaches with explanatory variables; (iii) we evaluate the forecast interval coverage as well as the point forecast accuracy; (iv) we observe the effect of temporal aggregation on the forecasting accuracy; and (v) we consider the mean absolute scaled error as an alternative forecasting accuracy measure. We find that pure time series approaches provide more accurate forecasts for tourism data than models with explanatory variables. For seasonal data we implement three fully automated pure time series algorithms that generate accurate point forecasts, and two of these also produce forecast coverage probabilities which are satisfactorily close to the nominal rates. For annual data we find that Naïve forecasts are hard to beat.     

Density forecasting through disaggregation

In this paper, the revised expectations model (REM) is developed to incorporate economic agents’ price expectation formation effects. With this incorporation, two models, an aggregate one sector model and a disaggregated multi-sector model, are estimated and used in density forecasting of the US real GDP growth rate. The experiment shows that use of the disaggregated version of the model, which incorporates price expectation effects along with modern Bayesian MCMC estimation and prediction techniques, produces more precise density forecasts than those yielded by either an aggregate version or benchmark forecasting models.     

Parameter-efficient deep probabilistic forecasting

Probabilistic time series forecasting is crucial in many application domains, such as retail, ecommerce, finance, and biology. With the increasing availability of large volumes of data, a number of neural architectures have been proposed for this problem. In particular, Transformer-based methods achieve state-of-the-art performance on real-world benchmarks. However, these methods require a large number of parameters to be learned, which imposes high memory requirements on the computational resources for training such models. To address this problem, we introduce a novel bidirectional temporal convolutional network that requires an order of magnitude fewer parameters than a common Transformer-based approach. Our model combines two temporal convolutional networks: the first network encodes future covariates of the time series, whereas the second network encodes past observations and covariates. We jointly estimate the parameters of an output distribution via these two networks. Experiments on four real-world datasets show that our method performs on par with four state-of-the-art probabilistic forecasting methods, including a Transformer-based approach and WaveNet, on two point metrics (sMAPE and NRMSE) as well as on a set of range metrics (quantile loss percentiles) in the majority of cases. We also demonstrate that our method requires significantly fewer parameters than Transformer-based methods, which means that the model can be trained faster with significantly lower memory requirements, which as a consequence reduces the infrastructure cost for deploying these models.     

Fast and frugal forecasting

Simple statistical forecasting rules, which are usually simplifications of classical models, have been shown to make better predictions than more complex rules, especially when the future values of a criterion are highly uncertain. In this article, we provide evidence that some of the fast and frugal heuristics that people use intuitively are able to make forecasts that are as good as or better than those of knowledge-intensive procedures. We draw from research on the adaptive toolbox and ecological rationality to demonstrate the power of using intuitive heuristics for forecasting in various domains including sport, business, and crime.     

On forecasting of innovations

The forecasting of basic technological innovations is focused. Clarification of the involved key concepts and an analytical framework are given. The forest industry serves as a background and illustration to the discussion. Considerable uncertainty prevails with respect to strategic investments in production in this industry. Industries, technological products, etc. have empirically been shown to pass similar phases of development over time. This life cycle starts with a basic innovation following on a previous invention. Due to competetion from more recent basic innovations the final phase is one of decline. Basic innovations are suggested to be exogenously determined and to cause structural change. The choice of forecasting strategy is suggested to be made dependent on the type of innovation, knowledge base, and life cycle position.     

Long-term forecasting and evaluation

Looking ahead thirty years is a difficult task, but is not impossible. In this paper we illustrate how to evaluate such long-term forecasts. Long-term forecasting is likely to be dominated by trend curves, particularly the simple linear and exponential trends. However, there will certainly be breaks in their parameter values at some unknown points, so that eventually the forecasts will be unsatisfactory. We investigate whether or not simple methods of long-run forecasting can ever be successful, after one takes into account the uncertainty level associated with the forecasts.     

Book Note

Book reviewed:
Andrew Savchenko, Rationality, Nationalism and Post-Communist Market Transformations: A Comparative Analysis of Belarus, Poland and the Baltic