Long memory cointegration and dynamic connectedness of volatility in US dollar exchange rates, with FOREX portfolio investment strategy

Abstract

Decisions of central banks on foreign exchange rates are based on the comovement of foreign exchange (FOREX) in mature markets such as US dollar rates to the British pound, euro, Chinese yuan, Japanese yen and Australian dollar. We investigate the long-run movement and dynamic quantile connectedness of volatility among pairs of these exchange rates. The updated residual-based fractional cointegration testing framework using narrow-band frequency domain least squares estimator is used to obtain the residual series for fractional cointegration. Quantile dynamic connectedness framework for volatility spillovers at different market conditions, depicted by quantiles, are used. We find evidence of long memory cointegration in seven pairs of exchange rates involving the previously mentioned currencies. These seven cases also correspond to a higher average index of quantile connectedness, with the effect of connectedness phasing out at higher quantiles and being more visible at lower quantiles. A portfolio investment strategy using optimal portfolio weights and hedge ratios for maintaining the accrued profit at the FOREX market is also presented.

Keywords

• exchange rate volatility
• narrow-band frequency domain least squares
• long memory cointegration
• quantile connectedness
• portfolio investment strategy