Characteristics of the Austrian passenger transport
There is a long-term multimodal eqilibrium given the Austrian Transport policy. Data about the number of cars, train passengers, air passengers, train passenger kilometers are used. Vector error correction models are utilized to show Granger short term and Johansen long-term equations. Data sources: (1) The number of personal cars and train passenger and passenger kilometers in thousands Main source: “Österreichische Verkehrsstatistik 19xx” issued by Statistik Austria. Data spikes and outliers compared to Statistisches Handbuch für die Republik Österreich (Österreichisches Statistisches Zentralamt, 1950-1991), Die österreichische Verkehrswirtschaft in Zahlen. Informationen der Bundessparte Transport und Verkehr der Wirtschaftskammer Österreich (Wirtschaftskammer Österreich, 2011-2020), Zahlen Daten Fakten (Österreichische Bundesbahnen, 2007-2021), Statistik Straße und Verkehr (Bundesministerium für Klimaschutz, Umwelt, Energie, Mobilität, Innovation und Technologie, 2000-2018); EU transport in figures. Statistical pocketbook (European Union, 2011-2018); Panorama of transport. Statistical overview of transport in the European Union. Data 1970-2001 (European Union, 2013). Car time series is without significant breaks or spikes and the methodology of counting (registered cars) is consistent. The methodology of counting the performance in “passenger-kilometers” changed in 2006 to the international standard, however, there is a long-term consistency in sampling and estimating techniques and the break-in series is not causing unexpected data variability. (2) The number of air transport passengers Main source: the World Bank database. This data covers the air traffic carried on scheduled services. However, air transport regulations in Europe have made it more difficult to classify air traffic as scheduled or nonscheduled. This time series is reported by International Civil Aviation Organization (ICAO) and represents the international and domestic scheduled traffic carried by the air carriers registered in a country.
Steps to reproduce
STATA code respected full VECM pre-estimation procedure: 1) vecrank to estimate the cointegrating rank of a VECM 2) varsoc to obtain lag-order selection statistics for VECMs Estimation procedure of the full model: 3) vec to estimate of 2 and 3 variate VECMS with robust standard errors (option vce(robust)) Only lags suggested by varsoc results. All trends estimated (C, RC, T, RT, None) - selected models with appropriate EC# lamdas (negative EC1 and positive EC2 combinations; three variables with at least negative EC1). Post estimation procedure: 4) veclmar to perform LM test for residual autocorrelation after vec 5) vecnorm to test for normally distributed disturbances after vec and in case of trend model a Shapiro-Wilk test of estimated residuals 6) vecstable to check the stability condition of VECM estimates Prais-Winsten and Cochrane-Orcutt regressions A standard estimation with the option vce(robust) Rank correlations using Spearman's correlations "spearman".