Economic uncertainty and exchange rates linkage revisited: modelling tail dependence with high frequency data

Battle Count: 80

The paper provides actionable insights for FX quantitative trading, particularly in identifying safe-haven currencies (CNY, BRL) during high uncertainty episodes. The time-varying tail dependence structure can inform dynamic portfolio allocation and hedging strategies. However, the paper does not provide direct trading signals, backtesting results, or transaction cost analysis. The findings are more relevant for strategic asset allocation and risk management than for high-frequency algorithmic trading. The identification of Gumbel copula (upper tail dependence) for CNY and BRL suggests specific regime-dependent trading opportunities.

Implementation requires: (1) ARMA-EGARCH estimation with GED distribution for each time series (requires specialized econometric software like RATS, EViews, or Python/R packages); (2) Time-varying copula estimation following Patton (2006) with logistic transformations and forcing variables; (3) Model selection across five copula families in both static and dynamic specifications; (4) Proper handling of marginal distribution transformations to uniform scores. The time-varying copula estimation is computationally intensive and requires careful initialization. No code is provided, and the specific parameterization details (e.g., exact ARMA orders selected) are not fully transparent.

Data sources are identified (TEU from policyuncertainty.com, exchange rates from DataStream), and the econometric methodology is well-described with equations. However, ARMA-EGARCH estimates are not reported (available upon request), and no code or repository is provided. The copula estimation procedure follows standard Patton (2006) framework, but specific software and parameter initialization details are missing.