nep-for New Economics Papers
on Forecasting
Issue of 2019‒04‒15
three papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting with a Global VAR model By Thomas van Florenstein Mulder; Tugrul Vehbi
  2. Time, space and hedonic prediction accuracy evidence from the Corsican apartment market By Yuheng Ling
  3. Feature Engineering for Mid-Price Prediction Forecasting with Deep Learning By Adamantios Ntakaris; Giorgio Mirone; Juho Kanniainen; Moncef Gabbouj; Alexandros Iosifidis

  1. By: Thomas van Florenstein Mulder; Tugrul Vehbi (Reserve Bank of New Zealand)
    Abstract: The Bank assesses the impact of international conditions on the New Zealand economy using a range of models. Among them, is the Global Vector Autoregressive model (GVAR), which is designed to analyse economic and financial interdependencies between countries. The GVAR is primarily used by the Bank to examine the transmission of global shocks or disturbances to the New Zealand economy. This Analytical Note examines to what extent the GVAR can also forecast macroeconomic conditions in New Zealand and its main trading partners. We test several specifications of the GVAR and calculate out-of-sample forecasts for GDP, inflation, interest rates, exchange rates and equity prices for New Zealand, U.S., China, Australia, Canada and the euro area. We then evaluate whether GVAR forecasts are more accurate than other statistical models and whether the model’s GDP forecasts can outperform economists’ forecasts published by Consensus Economics. We find that forecasts obtained from simple specifications of the GVAR tend to outperform other simple statistical models of inflation and GDP. The GVAR also outperforms economists’ GDP growth forecasts from Consensus Economics. These results emphasise the benefits of incorporating international linkages to improve forecast accuracy and suggest that the GVAR is a useful addition to the range of models used by the Reserve Bank to forecast the international economy.
    Date: 2019–03
  2. By: Yuheng Ling (Laboratoire Lieux, Identités, eSpaces et Activités (LISA))
    Abstract: In this study, we propose a hedonic housing model to address spa- tial and temporal latent structures simultaneously. With the development of spatial econometrics and spatial statistics, economists can now better assess the impact of spatial correlation on house prices. How- ever, the simultaneous handling of spatial and temporal correlation is still under development. Since spatial econometric models are limited to account for two kinds of cor- relation simultaneously, we propose using a hierarchical spatiotemporal model from spatial statistics. Based on a Bayesian framework and a stochastic par- tial differential equation (SPDE) approach, the estimation is carried out via INLA. We then perform an empirical study on apartment transaction prices in Corsica (France) using the proposed model. The empirical results demonstrate that the prediction performance of the hierarchical spatiotemporal model is the best among all candidate models. Moreover, the hedonic housing estimates are affected by spatial effects and temporal effects. Ignoring these effects could result in serious forecasting issues.
    Keywords: Hierarchical spatiotemporal model · Hedonic price model · INLA-SPDE · Apartment market
    JEL: C11 C14 C33 R31
    Date: 2019–04
  3. By: Adamantios Ntakaris; Giorgio Mirone; Juho Kanniainen; Moncef Gabbouj; Alexandros Iosifidis
    Abstract: Mid-price movement prediction based on limit order book (LOB) data is a challenging task due to the complexity and dynamics of the LOB. So far, there have been very limited attempts for extracting relevant features based on LOB data. In this paper, we address this problem by designing a new set of handcrafted features and performing an extensive experimental evaluation on both liquid and illiquid stocks. More specifically, we implement a new set of econometrical features that capture statistical properties of the underlying securities for the task of mid-price prediction. Moreover, we develop a new experimental protocol for online learning that treats the task as a multi-objective optimization problem and predicts i) the direction of the next price movement and ii) the number of order book events that occur until the change takes place. In order to predict the mid-price movement, the features are fed into nine different deep learning models based on multi-layer perceptrons (MLP), convolutional neural networks (CNN) and long short-term memory (LSTM) neural networks. The performance of the proposed method is then evaluated on liquid and illiquid stocks, which are based on TotalView-ITCH US and Nordic stocks, respectively. For some stocks, results suggest that the correct choice of a feature set and a model can lead to the successful prediction of how long it takes to have a stock price movement.
    Date: 2019–04

This nep-for issue is ©2019 by Rob J Hyndman. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.