Regional and Sectoral Business Cycles - Key Features for the Austrian economy

There exists a wealth of literature dealing with the analysis of business cycles, dating their turning points, measuring their synchronisation and identifying potential sources behind cyclical asymmetries, but most work concentrates on the countrywide level. The analysis of business cycle properties and dynamics at the regional level gained some popularity in the empirical literature, at least in the European context, fostered by the implementation of the single monetary market within Europe and its ongoing economic and monetary integration process. This strand of literature mainly deals with the question whether regional cycles across European countries have become more or less synchronised since the start of the currency union. However, most of these studies use NUTS level 1 data in their level of regional analysis, which in the case of Austria relates only to three regional areas (East/South/West) - a drawback, if one is interested in business cycle features on a more disaggregated regional level (e.g. NUTS 2 - federal districts). So far, to my best knowledge, no empirical investigation focusing on regional business cycles in the Austrian economy beyond NUTS level 1 has been conducted. The aim of this paper is to fill this gap and provide stylized facts for the nine Austrian federal districts as well as for the various sectors (e.g. manufacturing, construction, services etc.) in each region with respect to their cyclical properties and synchronisation dynamics over time and to contrast them with the national aggregates of the business cycle.The dataset used for analysing regional business cycles in the Austrian economy contains quarterly estimates of real gross value added (GVA). Data for the nine Austrian federal provinces (on a sectorial basis and in aggregation) are available back to 1988. The regional statistical data for the Austrian federal provinces correspond to the NUTS level 2 classification established by Eurostat. The modeling approach in this study is as follows: First, I derive the business cycle component following the classical approach (expansions and contractions in the level of output) as well as the deviation cycle framework (periods of above-trend and below-trend rates of economic growth). Second, I establish for the various methods at hand a business cycle chronology for each region (and sector) and contrast their cyclical properties (e.g. duration and amplitude). For the subsequent analysis of business cycle synchronisation I use Corbae-Ouliaris filtered data and turning points obtained with the Bry-Boschan dating routine as input and investigate the degree of synchronisation employing cross-correlation, coherence and concordance measures. In applying the Corbae-Ouliaris (2003) filter I take advantage of the superior statistical properties of this filter compared to the popular Baxter-King and Hodrick-Prescott filters and, thus, expect more robust results. By using frequency domain techniques and recent developments in spectral regression for non-stationary time series the authors proposed an approximate ‘ideal’ band pass filter and showed that the new filter does not exhibit phase-shifts at the end of the filtered series and reduces leakages around the cut-off frequencies.Preliminary results on cyclical synchronisation show that the regional cycles are quite heterogeneous and that the degree of regional co-movement with the national business cycle is rather weak at the beginning of the sample. However, the cyclical conformity increases substantially for most of the regions from around 2000 onwards. On a provinces scale, Upper Austria and Vorarlberg exhibit the most consistent synchronised movement with the Austrian business cycle. Burgenland, in contrast, shows the least conformity. These very preliminary findings are in line with results derived in studies using higher aggregated regional data (e.g. NUTS level 1), but the degree of regional convergence amongst the Austrian federal districts is not as pronounced as in the aggregated case.