Identifying Turkish business cycle regimes in real time

ABSTRACT

In this study, we analyse the real-time identification performance of the BBQ method and the Markov switching (MS) model in the case of Turkey by comparing their real-time and ex-post identification results between 1997M01-2017M12. We show that both the BBQ and the MS methodologies identify the nearly same turning point dates for the Turkish economy both ex-post and in real time by using a pseudo real-time data set. We also calculate the real-time identification lag of models and show that the MS model and the BBQ method identify a turning point with a 3–4 months lag and a 6 months lag, respectively. Finally, we show that data revisions do not have a significant impact on the real-time identification performance of the models between 2005M01-2017M12.

KEYWORDS:

• Business cycles
• turning points
• Markov switching
• BBQ
• Turkey

JEL CLASSIFICATION:

• E32
• E37

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ There is no long-term vintage data set for Turkey.

² We observe that the MS model with an AR component produces noisy probabilities with numerous false cycles. On the other hand, the MS model without an AR component produces business cycle probabilities without false cycles. Ferrara (2003) also uses an MS model without an AR component for classifying business cycle states in the United States.

³ Changing the threshold has negligible effect on the MS model and no effect on the BBQ model as the BBQ model classifies business cycle states either 1 or 0.

⁴ There are only a few macroeconomic variables with long time span and among them, the import volume index and the real sector confidence index are the best leading variables for the IPI.

⁵ For real-time identification of turning points, this is a robust method and different variants of this method used frequently in the literature (e.g. Chauvet and Piger 2008; Giusto and Piger 2017; Huang and Startz 2018; Soybilgen 2018).

⁶ Decreasing or increasing this lag has no significant impact on results.

⁷ For example, the BBQ method establishes the start of the 1998–1999 recession as September of 1998 in March of 1999. Therefore, the identification lag is 6 months.

⁸ Due to lack of data, it is not possible to extend the vintage data set more than this.