A study on correlations in financial market

Abstract

In this thesis, a local linear method is proposed to study the correlations between financial assets (markets) during different market conditions. The impact of market return and market volatility on correlation is of main interest. The asymptotic properties of the estimator are investigated and expressions of the confidence band are derived. In real data test, equity correlations are shown to increase sharply during strong bear/bull market while market volatility has little influence on it. Furthermore, the possible lead-lag structure in equity return is examined with high frequency returns