Abstract
We analyze the cyclical dynamics of the Turkish economy and the stock market as well as their interactions. We use hidden Markov models that are robust to parameter instability arising from major shifts in economic policy, which have been typically observed in the Turkish economy. These models provide estimates of turning points for the growth, business, and stock market cycles. We identify three states of growth cycles and two states of business cycles in Turkey characterized by different mean estimates. We find that the economy went through five recessions since 1987. Crises are characterized by sharp drops in economic activity and are preceded by slowdowns. These crises are typically followed by strong recoveries during which the economy grows above its long-run average rate. We show that the Turkish stock market goes through three regimes having distinct risk-return dynamics. Bear markets associated with negative returns precede every recession with an average lead time of three quarters, suggesting that the stock market may be a useful forward-looking indicator of the Turkish economy.
Acknowledgements
The authors would like to thank participants of the 6th Eurostat Colloquium on Modern Tools for Business Cycle Analysis for useful comments. The views expressed in this paper are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System, its staff or any other person associated with the Federal Reserve System.
Notes
1According to the Economist (Economics Focus, December 10, 2011) Turkey is one of the growth markets, defined as an emerging economy that accounts for more than 1% of global GDP, alongside the BRICS, Indonesia, Mexico, and South Korea.
2See Rodrik (Citation2012) for a detailed assessment of the recent performance of the Turkish economy.
3For studies that focus on US, see Perez-Quiros and Timmermann (Citation1995), Hamilton and Lin (Citation1996), Chauvet (Citation1998/1999), Senyuz (Citation2011) among others. For examples of studies that analyze linkages between the stock market and real economy for a group of other advanced or emerging countries, see Candelon and Metiu (Citation2011), Cheung and Ng (Citation1998), and Maysami and Sim (Citation2001).
4This is in contrast with ad hoc non-parametric rules that require a substantial amount of ex-post data. One example is the Bry and Boschan (Citation1971) algorithm that attempted to formalize the official NBER dating rules.
5Note that this is the general form of the model. Under the constant variance assumption, the model boils down to a mean-switching only specification.
6See Chauvet and Potter (Citation2001), Kim and Nelson (Citation1999), and Koop and Potter (Citation2000), that also document this break.
7See Chauvet (Citation2002) for an application on Brazilian economy.
8See Hansen (Citation1992) for another testing procedure where the supremum of the calculated standardized LR statistics is utilized.
9See Chauvet (Citation1998/1999) for a similar approach in relating stock market dynamics to business cycles.