nep-for New Economics Papers
on Forecasting
Issue of 2018‒02‒12
four papers chosen by
Rob J Hyndman
Monash University

  1. Modelling and Forecasting Tourist Arrivals to Cambodia: An Application of ARIMA-GARCH Approach By Chhorn, Theara; Chaiboonsri, Chukiat
  2. Evaluating Medium Term Forecasting Methods and their Implications for EU Output Gap Calculations By Kieran Mc Morrow; Werner Roeger; Valerie Vandermeulen
  3. Nonfractional Memory: Filtering, Antipersistence, and Forecasting By J. Eduardo Vera-Vald\'es
  4. A short-term forecasting model for the Spanish economy: GDP and its demand components By Ana Arencibia Pareja; Ana Gómez Loscos; Mercedes de Luis López; Gabriel Pérez Quirós

  1. By: Chhorn, Theara; Chaiboonsri, Chukiat
    Abstract: The paper aims at estimating and forecasting international tourist arrivals to Cambodia during the time interval of 2000m1 to 2017m7, covering 209 of monthly observations. To find out factors affecting tourist arrivals, simple OLS and 2SLS with instrument variable regression are applied, on the one hand. On the other hand, several time series models of ARIMA (p, d, q), GARCH (s, r) and the hybrid of ARIMA(p, d, q)-GARCH(s, r) are employed to forecast tourist arrivals in line with AIC and BIC in selecting the best modified models. The empirical results primarily reveal that tourist arrivals are affected by exogenous factor, say exchange rate, dummy factors such as the AEC, global finical crisis, national election and Cambodia’s e-Visa. With regard to forecasting stage, the result indicates that tourist arrivals are shocked by time trend in the past period, say time (t-1). The trend is furthermore reduced due to the time lags, say time (t-2, t-3) as shown in the parameter coefficients of AR. GARCH (1, 1) model suggests that the short run persistence of shocks lies in the gap of 0.04 whereas the long run persistence lies in the gap of 0.94. Additionally, AIC and BIC propose that ARIMA(3, 1, 4) and the hybrid of ARIMA(3, 1, 4)-GARCH (1, 1) are the best model to predict the future value of tourist arrivals. The RMSE and U index obtained from measurement predictive accuracy reveal that long run 1-step ahead forecasting of 2013m12 to 2017m7 is produced the smallest predictive error amongst the others. Thus, it has more predictive power to apply long term ex-ante forecasting.
    Keywords: Point Forecasting Interval, out of Sample Forecasting, ARIMA (p, d, q)- GARCH (s, r) Model, Exchange rate and Dummy Factors, Tourist Arrivals, Cambodia
    JEL: C22 C53
    Date: 2017–12–13
  2. By: Kieran Mc Morrow; Werner Roeger; Valerie Vandermeulen
    Abstract: This paper sheds light on two specific, but interlinked, questions – firstly, how do the EU's, medium term actual GDP growth rate forecasts compare, in terms of accuracy and biasedness, with those of the EU's Member States, in their annual Stability and Convergence Programme (SCP) updates; and secondly, should medium term forecasts be allowed to influence the short run output gap and structural balance calculations used in the EU’s fiscal surveillance procedures. Regarding the first question, the paper concludes that the EU's medium term forecasts are equally as good, and arguably better, than those of the SCP's both with respect to accuracy and biasedness. Regarding the second question, due to the relatively rapid loss in forecast accuracy as the time horizon lengthens; the paper suggests that using more forecast information should be avoided in the output gap and structural balance calculations. Extending the forecast horizon to be used in the output gap calculations could exacerbate an existing optimistic bias with respect to the supply side health of the EU’s economy, thereby enlarging the risk of procyclicality problems, especially in the upswing phase of cycles, where most of the large fiscal policy errors tend to occur.
    JEL: C10 E60 O10
    Date: 2017–10
  3. By: J. Eduardo Vera-Vald\'es
    Abstract: The fractional difference operator remains to be the most popular mechanism to generate long memory due to the existence of efficient algorithms for their simulation and forecasting. Nonetheless, there is no theoretical argument linking the fractional difference operator with the presence of long memory in real data. In this regard, one of the most predominant theoretical explanations for the presence of long memory is cross-sectional aggregation of persistent micro units. Yet, the type of processes obtained by cross-sectional aggregation differs from the one due to fractional differencing. Thus, this paper develops fast algorithms to generate and forecast long memory by cross-sectional aggregation. Moreover, it is shown that the antipersistent phenomenon that arises for negative degrees of memory in the fractional difference literature is not present for cross-sectionally aggregated processes. Pointedly, while the autocorrelations for the fractional difference operator are negative for negative degrees of memory by construction, this restriction does not apply to the cross-sectional aggregated scheme. We show that this has implications for long memory tests in the frequency domain, which will be misspecified for cross-sectionally aggregated processes with negative degrees of memory. Finally, we assess the forecast performance of high-order $AR$ and $ARFIMA$ models when the long memory series are generated by cross-sectional aggregation. Our results are of interest to practitioners developing forecasts of long memory variables like inflation, volatility, and climate data, where aggregation may be the source of long memory.
    Date: 2018–01
  4. By: Ana Arencibia Pareja (Banco de España); Ana Gómez Loscos (Banco de España); Mercedes de Luis López (Banco de España); Gabriel Pérez Quirós (Banco de España)
    Abstract: This document describes the key aspects of the extended and revised version of Spain-STING (Spain, Short-Term Indicator of Growth), which is a tool used by the Banco de España for the short-term forecasting of the Spanish economy’s GDP and its demand components. Drawing on a broad set of indicators, several dynamic factor models are estimated. These models allow the forecasting of GDP, private consumption, public expenditure, investment in capital goods, construction investment, exports and imports in a consistent way. We assess the predictive power of the GDP and its demand components for the period 2005- 2017. With regard to the GDP forecast, we find a slight improvement on the previous version of Spain-STING. As for the demand components, we show that our proposal is better than other possible time series models.
    Keywords: business cycles, spanish economy, dynamic factor models.
    JEL: E32 C22 E27
    Date: 2018–02

This nep-for issue is ©2018 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.