Seyfettin Gürsel, Ozan Bakış and Uğurcan Acar
TurkStat started to estimate national accounts using a new method (chained volume index) and new data sets (especially firm tax records) from third quarter of 2016 that caused some controversies. Aside from the problems revealed by these controversies it seems that an additional problem emerged nowadays. We are witnessing large and sometimes opposite revisions in the annually and quarterly growth estimates released by the TurkStat. It is a common practice to make some revisions in each new statement after the first estimate of GDP. In order to release growth estimates as early as possible the whole collection of the related data is not considered as a necessity. Once the whole data is completed, the estimates announced previously are revised by national statistical institutes. However, these revisions remain within reasonable limits and are damped after a certain period of time.
However, when press releases by TurkStat on five quarters between third quarter of 2016 and third quarter of 2017 are examined, it is observed that the revisions made by TurkStat in the national accounts might hardly be considered as reasonable since revisions, often large and even in opposite directions, follow each other, and the expected damping does not occur in the repeated revisions. It is easy to admit that these revisions point to an exceptional case when these revisions are compared with the revisions made by TurkStat in old series and with the practices of the statistical institutes of Germany and France. We think that large and opposite revisions in past growth estimates create problems in understanding the dynamics of Turkish economy and make difficult to forecast future growth rates.