We propose a vine copula model based on a bivariate extended skew-t distribution and derive its corresponding multivariate tail dependence function. Our simulations demonstrate that the proposed estimator dominates the conventional vine copula approach in the estimation of multivariate tail dependence. We apply our model to a safe haven analysis of US dollars (US$) and gold prices against stocks. The estimated multivariate lower tail dependence coefficients suggest that even though either US$ or gold can be safe haven assets against stocks, combining US$ and gold in a portfolio does not provide a safe haven property against stocks. Therefore, incor- porating multiple safe haven assets in a portfolio may end in heavier losses in the event of a market downturn. Our results highlight the importance of simultaneously investigating multiple safe haven assets in financial risk analysis.