Abstract
The Asia-Pacific region remains understudied regarding return connectedness among exchange rate markets. Furthermore, previous studies mostly use average-based estimators to measure return connectedness, even though they are not necessarily suitable to quantify connectedness measures during extreme events, i.e. in the tails of the conditional distribution. The aim of this paper is to study the dynamic connectedness among 11 Asia-Pacific exchange rates from September 1994 to August 2019, not only on average but also in the left and right tails. Using mean-based measures of connectedness, the level of return connectedness is on average 32%, indicating a moderate level of connectedness across the currencies under study. However, using a novel quantile-based connectedness framework, the level of connectedness is found to be much higher, reaching around 85% at both the left and right tails of the conditional distribution. This result shows that return connectedness across the 11 exchange rates is much stronger during extreme negative and extreme positive events, suggesting that the application of mean-based models of connectedness in the Asia-Pacific exchange rate markets is too restrictive and inadequate. Furthermore, return connectedness measures are time-varying in all cases, but are less volatile in the tails. A detailed analysis involving the relative tail dependence shows that an asymmetric behaviour is present, indicating that return spillovers differ between periods of extreme market depreciations and periods of extreme market appreciations.
•Study extreme spillovers across Asia-Pacific currencies via quantile-based analyses.•Applying mean-based models of connectedness is too restrictive and inadequate.•Return spillovers are stronger at the tails of the distribution than at the mean.•Total return connectedness index is time-varying but is less volatile at the tails.•Relative tail dependence analysis shows evidence of an asymmetric behaviour.