Gold is often considered a safe haven asset providing negative return correlation with the stock market in times of distress, while in more calm periods the correlation is close to zero. We study the dynamic inter-linkage of gold prices and the stock market. Specifically, we model the log-prices of gold and a stock index as jump-diffusive processes, with the jumps arriving with mutually exciting intensities. Hence, the occurrence of negative shocks to the stock index spill over into higher probabilities of positive shocks to the gold price and vice versa. For the empirical analysis, we consider daily prices on gold and the SPX index. Utilizing that the model's moment conditions are computed efficiently in closed form, we use the generalized method of moments to estimate the model parameters. We document the existence of cross-excitation between the stock index and gold prices, with the channel from the stock index to gold prices being the most pronounced. Moreover, we find that gold behaves as a safe haven asset for the stock index for around 20 days following a market crash. Finally, we study the power of the proposed model to predict future price jumps and benchmark the performance against more classical models.