Dinamik faktör modelleri ve finansal bağlanmışlık: Dünyadaki ulusal bankacılık sistemleri üzerine bir uygulama

This paper analyzes the volatility connectedness of major banking systems around the world. Incorporating dynamic factor models into Diebold-Yılmaz connectedness framework (Diebold and Yilmaz, 2011), we calculate volatility connectedness measures for countries rather than individual banks. In each country, bank stock return volatilities are assumed to be driven by a common country factor and an idiosyncratic component unique to each bank. The common country factor is calculated as the first principal component of the bank stock return volatilities. Constructing a VAR model of the country volatility factors, we undertake the variance decomposition analysis of volatility shocks to obtain the Diebold-Yılmaz connectedness measures. We obtain both static and dynamic measures of connectedness. In the static analysis we show that the volatility connectedness of countries are closely linked to major banking system characteristics. First, banking systems located in the same region tend to be more connected with each other in volatility than with those outside the region. Second, country banking systems tend to generate net volatility connectedness towards others as their size and financial development levels rise. In the dynamic rolling window analysis, we obtain important results related to both systemic volatility and volatility transmitted by individual countries. In terms of systemic risk, we manage to capture the main stages of the crisis by our total connectedness index. We find that the US was the main generator of volatility to other countries from the onset of the US financial crisis until the end of 2008. However, once the crisis became global and was followed by the sovereign debt/banking crisis in the Eurozone periphery, the "to-connectedness" of the European banking system, and especially the ones in the southern periphery, increased substantially. Keywords: Financial connectedness, dynamic factor models, risk measurement,systemic risk, systemically important financial institutions, vector autoregression,variance decomposition