Search results

In the last few decades, there has been growing interest in forecasting with computer intelligence, and both fuzzy time series (FTS) and artificial neural networks (ANNs) have gained particular popularity, among others. Rather than the conventional methods (e.g., econometrics), FTS and ANN are usually thought to be immune to fundamental concepts such as stationarity, theoretical causality, post-sample control, among others. On the other hand, a number of studies significantly indicated that these fundamental controls are required in terms of the theory of forecasting, and even application of such essential procedures substantially improves the forecasting accuracy. The aim of this paper is to fill the existing gap on modeling and forecasting in the FTS and ANN methods and figure out the fundamental concepts in a comprehensive work through merits and common failures in the literature. In addition to these merits, this paper may also be a guideline for eliminating unethical empirical settings in the forecasting studies.

Retail sales usually exhibit strong trend and seasonal patterns. Practitioners have typically used seasonal autoregressive integrated moving average (ARIMA) models to predict retail sales exhibiting these patterns. Due to economic instability, recent retail sales time-series data show a higher degree of variability and nonlinearity, which makes the ARIMA model less accurate. This chapter demonstrates the feasibility and potential of applying empirical mode decomposition (EMD) in forecasting aggregate retail sales. The hybrid forecasting method of integrating EMD and neural network (EMD-NN) models was applied to two real data sets from two different time periods. The one-period ahead forecasts for both time periods show that EMD-NN outperforms the classical NN model and seasonal ARIMA. In addition, the findings also indicate that EMD-NN can significantly improve forecasting performance during the periods in which macroeconomic conditions are more volatile.

This chapter extends the univariate forecasting method proposed by Wang, Luc, and Hsiao (2013) to forecast the multivariate long memory model subject to structural breaks. The approach does not need to estimate the parameters of this multivariate system nor need to detect the structural breaks. The only procedure is to employ a VAR(k) model to approximate the multivariate long memory model subject to structural breaks. Therefore, this approach reduces the computational burden substantially and also avoids estimation of the parameters of the multivariate long memory model, which can lead to poor forecasting performance. Moreover, when there are multiple breaks, when the breaks occur close to the end of the sample or when the breaks occur at different locations for the time series in the system, our VAR approximation approach solves the issue of spurious breaks in finite samples, even though the exact orders of the multivariate long memory process are unknown. Insights from our theoretical analysis are confirmed by a set of Monte Carlo experiments, through which we demonstrate that our approach provides a substantial improvement over existing multivariate prediction methods. Finally, an empirical application to the multivariate realized volatility illustrates the usefulness of our forecasting procedure.

Expectations play a crucial role in finance, macroeconomics, monetary economics, and fiscal policy. In the last decade a rapidly increasing number of laboratory experiments have been performed to study individual expectation formation, the interactions of individual forecasting rules, and the aggregate macro behavior they co-create. The aim of this article is to provide a comprehensive literature survey on laboratory experiments on expectations in macroeconomics and finance. In particular, we discuss the extent to which expectations are rational or may be described by simple forecasting heuristics, at the individual as well as the aggregate level.

This is a study of forecasting models that aggregate monthly times series into bimonthly and quarterly models using the 1,428 seasonal monthly series of the M3 competition of Makridakis and Hibon (2000). These aggregating models are used to answer the question of whether aggregation models of monthly time series significantly improve forecast accuracy. Through aggregation, the forecast mean absolute deviations (MADs) and mean absolute percent errors (MAPEs) were found to be statistically significantly lower at a 0.001 level of significance. In addition, the ratio of the forecast MAD to the best forecast model MAD was reduced from 1.066 to 1.0584. While those appear to be modest improvements, a reduction in the MAD affects a forecasting horizon of 18 months for 1,428 time series, thus the absolute deviations of 25,704 forecasts (i.e., 18*1,428 series) were reduced. Similar improvements were found for the symmetric MAPE.

There is a growing interest in fuzzy time series (FTS) forecasting, and several improvements are presented in the last few decades. Among these improvements, the development of causal models (i.e., multiple factor FTS) has sparked a particular literature dealing with the causal inference and its integration in the FTS framework. However, causality among variables is usually introduced as a subjective assumption rather than empirical evidence. As a result of arbitrary causal modeling, the existing multiple factor FTS models are developed with implicit forecasting failure. Since post-sample control (unknown future, as in the business practice) is usually ignored, the spurious accuracy gain through increasing factors is not identified by scholars. This paper discloses the use of causality in the FTS method, and investigates the spurious causal inference problem in the literature with a justification approach. It invalidates the contribution of dozens of previously published papers while justifying its claim with illustrative examples and a comprehensive set of experiments with random data, as well as real business data from maritime transportation (Baltic Dry Index).

We conduct a detailed simulation study of the forecasting performance of diffusion index-based methods in short samples with structural change. We consider several data generation processes, to mimic different types of structural change, and compare the relative forecasting performance of factor models and more traditional time series methods. We find that changes in the loading structure of the factors into the variables of interest are extremely important in determining the performance of factor models. We complement the analysis with an empirical evaluation of forecasts for the key macroeconomic variables of the Euro area and Slovenia, for which relatively short samples are officially available and structural changes are likely. The results are coherent with the findings of the simulation exercise and confirm the relatively good performance of factor-based forecasts in short samples with structural change.

This research examines the use of a number of time series model structures of a moderate allocation mutual fund, PRWCX. PRWCX was rated as the top fund in its category during the past five years. The fund invests at least 50% of its total assets that the fund manager believes that have above average potential for capital growth. The remaining assets are generally invested in convertible securities, corporate and government debt bank loans, and foreign securities. Forecasting the total NAV of such a moderate allocation mutual fund, composed of an extremely large number of investments, requires a method that produces accurate results. These models are exponentially smoothing (single, double, and Winter’s Method), trend models (linear, quadratic, and exponential) are Box-Jenkins models.

Machine learning methods have recently gained attention in business applications. We will explore the suitability of machine learning methods, particularly support vector regression (SVR) and radial basis function (RBF) approximation, in forecasting company sales. We compare the one-step-ahead forecast accuracy of these machine learning methods with traditional statistical forecasting techniques such as moving average (MA), exponential smoothing, and linear and quadratic trend regression on quarterly sales data of 43 Fortune 500 companies. Moreover, we implement an additive seasonal adjustment procedure on the quarterly sales data of 28 of the Fortune 500 companies whose time series exhibited seasonality, referred to as the seasonal group. Furthermore, we prove a mathematical property of this seasonal adjustment procedure that is useful in interpreting the resulting time series model. Our results show that the Gaussian form of a moving RBF model, with or without seasonal adjustment, is a promising method for forecasting company sales. In particular, the moving RBF-Gaussian model with seasonal adjustment yields generally better mean absolute percentage error (MAPE) values than the other methods on the sales data of 28 companies in the seasonal group. In addition, it is competitive with single exponential smoothing and better than the other methods on the sales data of the other 15 companies in the non-seasonal group.