Capital flows and economic growth across spectral frequencies: Evidence from Turkey

In this paper we study the interactions and feedbacks between three categories of net capital flows and growth in the Turkish economy for the 1992:01-2009:01 period using frequency domain techniques. Our main spectral analysis tool is a new version of the causality test of Geweke (1982) and Hosoya (1991) in the frequency domain developed by Breitung and Candelon (2006). Besides, we make use of other tools of spectrum analysis such as cospectrum, squared coherence, phase and gain spectrums to decompose total covariance between capital flows and growth across main frequency bands and capture lead/lag interactions between them. Some of our empirical findings are as follows: Variance decompositions over frequency bands reveal that variations in individual capital flow categories are concentrated over high (seasonal) frequencies. We found no feedback from short-term and long-term 'other' investments to growth in these frequencies. However, there are highly significant feedbacks from growth to short-term and long-term capital inflows over business cycle and seasonal frequencies. Spectral variance decompositions reveal that, in general, percentage of variation in capital flows due to economic growth is much higher than the percentage of variation in growth due to capital flows.