In this blog post graphic methods are used to describe quarterly GDP time series in six countries.
Two types of a line graph could be used to show changes in quarterly GDP over time. The first displays both original and seasonally and calendar adjusted time series for Croatia, Serbia and Slovenia on a single graph. This graph shows the relative position of time series to each other. However, if the time series levels are disproportional in size then on this graph is difficult to spot patterns in time series.
The second graph displays each series on a separate panel. On such graph any persistent pattern related to trend or seasonal factors should be clearly visible. Figure 1 shows line graphs of quarterly GDP in Croatia, Serbia and Slovenia on the same panel.
A couple of features of these series stand out. First, all three series show changes in the level. After year 2000 (excluding Serbian data impacted by war, sanctions and NATO bombing of Federal Republic of Yugoslavia in 1999) all three series show very similar patterns. The rank list of these countries according to the total volume of GDP: Croatia, Serbia and Slovenia. There is a persistent upward trend in these series until 2007-2008 when the great financial crisis hit the region. Then there is a significant drop in the level with a bit different paths to recovery. GDP in Croatia shows negative, downward trend until 2015 and then the GDP series picked up. In case of Serbia drop in the level was not to such extent and series shows oscillation around some constant level until 2013 when the growth continue at the similar rate as in Croatia. Finally, in case of Slovenia, drop in the GDP level was mild and series keeps oscillating around some constant level until 2015 when the growth continue, but not at the same rate as in Croatia and Serbia.
Second, all three series show a persistent pattern of short-term volatility around the trend in the whole period. This short term volatility seems to be related to seasonal factors. However, these three series show quite distinct seasonal patterns. In case of Croatia there is an increasing short term volatility after the global financial crisis. In case of Slovenia seasonal pattern before and after the global financial crisis has changed. However, the GDP in Slovenia is the least volatile series among these three series. Finally, in case of Serbia the seasonal pattern in GDP is quite stable.
Figure 2 shows line graphs of quarterly GDP in Croatia, Serbia and Slovenia on the separate panels.
We have already commented on the long-term trend and seasonal patterns in these three series.
In Figure 3 all three series show long-term upward trend. However, seasonal patterns are quite different. While only the Bosnia and Herzegovina quarterly GDP series shows stable seasonal pattern, Macedonia and Montenegro quarterly GDP time series show great variations. This fact may indicate some kind of seasonal random walk behaviour. This will be further investigated using seasonal plots and decomposition of time series.