Milli Gelir Verilerindeki Guncelleme Sonrasi Kisa Donemli Tahmin Modellerinin Yenilenmesi

[TR] Turkiye Istatistik Kurumu’nun milli gelir verilerinde yaptigi guncelleme sonrasi iktisadi faaliyete dair hangi gostergenin takip edilmesi gerektigi ve tahmin modellerinde ne tur gostergelerin kullanilmasinin faydali olacagi konusu siklikla gundeme gelmektedir. Bu notta, soz konusu hususlari ele almak amaciyla yeni milli serisinin kisa donemli tahminine yonelik cesitli model sonuclari ozetlenmektedir. Calisma kapsaminda; reel gostergeler, finansal veriler, anket verileri ile fiyat ve butce istatistiklerine iliskin 379 farkli seriden meydana gelen bir veri seti olusturulmus ve bu verilerin yeni milli gelir serisini tahmin performansina gore en iyi sonucu veren modeller belirlenmistir. En iyi tahmin sonucunu veren modellerde vergi gelirleri ve tahsili gecikmis alacak oranlari gibi degiskenlerin yer aldigi bulgulanmistir. Eski milli gelir serisinin tahmin edildigi modellerde imalat sanayi uretimi basarili sonuc verirken, yeni milli gelir serisinin tahmininde imalat sanayi uretimi yerine sektorlerin ihracat paylarina gore agirliklandirilarak kullanilmasinin tahmin performansini iyilestiren bir etken oldugu kaydedilmistir. [EN] The revision of the national income series by the Turkish Statistical Institute brought up a question as to what kind of indicators should be used to monitor economic activity and which variables can be used in forecasting models. This study aims to address these issues by estimating various short-term forecasting models for the new national income series. Accordingly, a comprehensive data was formed including 379 series, which are comprised of real sector indicators, financial data, survey-based data as well as price and budget statistics, and the best models were selected on the basis of their forecasting performance. In this context, the best performing models include tax revenues and non-performing loans. Meanwhile, manufacturing industry production performed quite well in forecasting the old national income series, whereas sectoral data based on export share is more successful in forecasting the new income series.