New data sources in statistics: the effect of freight intensity on macroeconomic indicators

Peter Knížat, Statistical Office of the Slovak Republic, University of Economics in Bratislava, Slovak Republic
Dagmar Ceľuchová Bošanská, Martin Janík, Filip Nguyen, Alistiq, s. r. o., Slovak Republic

Type of article: scientific article
Pages: 7 – 26


The aim of this article is to verify the possibilities of using the SARIMA and SARIMAX models to estimate the development of Slovakia’s GDP. In the SARIMAX model, the data on kilometers travelled in freight transport were used as an additional exogenous variable. After decomposing the Slovak electronic toll system data into its component parts, seasonal component and enough residuals component were identified, indicating that the data time series is not stationary. In fitting the SARIMA/SARIMAX models, the Augmented Dickey-Fuller (ADF) test was used to check the time series stationarity and the Moving Average Method was used for the time series transformation. The adequacy of the fitted models was confirmed by observing the Ljung-Box test result. Moreover, the JDemetra+ software was utilized for seasonal analysis of time series. According to the stationarity test of the time series using the Akaike information criterion (AIC), the fitted SARIMAX model was the most suitable to forecast the GDP of Slovakia. The results from the Root Mean Squared Percentage Error (RMSPE) and the Mean Absolute Percentage Error (MAPE) indicate the superiority of the SARIMAX model, which took into account the seasonality patterns and exogenous factors. The SARIMAX model outperformed the SARIMA, predicting values within the 95 % confidence interval, with a RMSPE value of 8.9 % while the SARIMA had a RMSPE of 17.4 %. The conclusions of this article indicate that non-conventional data sources can have a high potential of use for the estimation of economic indicators, that can provide a more comprehensive and up-to-date view of the economic situation.

Issue for download
PDF (3.1 MB, 54 downloads)