Abstract
We construct a sovereign default network by employing high-dimensional vector autoregressions obtained by analyzing connectedness in sovereign credit default swap markets. We develop four measures of centrality, namely, degree, betweenness, closeness, and eigenvector centralities, to detect whether network properties drive the currency risk premia. We observe that closeness and betweenness centralities can negatively drive currency excess returns but do not exhibit a relationship with forward spread. Thus, our developed network centralities are independent of an unconditional carry trade risk factor. Based on our findings, we develop a trading strategy by taking a long position on peripheral countries' currencies and a short position on core countries' currencies. The aforementioned strategy generates a higher Sharpe ratio than the currency momentum strategy. Our proposed strategy is robust to foreign exchange regimes and the coronavirus disease 2019 pandemic.