- January 2021 (3)
- December 2020 (4)
- November 2020 (5)
- October 2020 (4)
- September 2020 (4)
- August 2020 (4)
- July 2020 (1)
- June 2020 (4)
- May 2020 (5)
- April 2020 (3)
- March 2020 (6)
- February 2020 (3)
Caution against over-information
I just want to stress that we have to be cautious and avoid over-excitation when using seasonally and working day adjusted series.
Last time I said over-information might be problematic for people who try to interpret the information. For instance, for industrial output, raw, seasonally adjusted, and seasonally and working day adjusted figures are released. In fact, it is good that the Turkish Statistical Institute, TURKSTAT, releases all three series. But if not handled with care, the three “states” of industrial output cause misinterpretations.
Change in industrial output in comparison to the previous year or month might be illuminating. But if the series are influenced by seasonal as well as economic changes, we cannot be sure which factor primarily caused the change. I will stop here, but there is a note for those who are interested in hearing further. From this perspective, seasonally and working day adjusted figures are useful in making comparisons between periods.
But one needs to be cautious when assessing changes between two periods: consistent increases or decreases in consecutive periods do not necessarily offer a clear change in the main trend. I will exemplify this using industrial output data which is critical for growth performance.
I will use quarterly figures in order to avoid monthly fluctuations. The first “state” of the series is the annual year-on-year percentage change in quarterly data. This is how I calculate the main trend in the series. The second “state” is the year-on-year percentage change in quarterly and seasonally and working day adjusted series.
The figure below shows the movements. Seasonally and working day adjusted figures increased in three consecutive quarters starting in the second quarter of 2011. That is, industrial output growth rate showed a gradual increase. This movement is denoted with an arrow on the figure. Nevertheless, it is wrong to interpret this as gradual economic buoyancy since the trend of industrial output growth weakened throughout the studied period as annual rates reveals. Temporary fluctuations in the adjusted series are also worth noting.
I am not telling that adjusted series should not be used. Do not get me wrong. I just want to stress that we have to be cautious and avoid over-excitation.
Note: Tourism revenues, for instance, increase in summers and decrease in winters. An increase in revenues in summer compared to winter that year does not have an economic significance. The increase might be a result of the usual summer effect even though the number of tourists and spending per tourist had not changed. The number of tourists and spending might have increased as well. Since we are concerned with the second option, we use adjusted data to eliminate the first mechanism.
This commentary was published in Radikal daily on 11.04.2013