Is there anyone willing to suggest a good way to estimate the jump parameters in the Merton jump model ?The parameters include jump rate, jump size mean and jump size volatility.I know there are people using EM to estimate those parameters.Any reply and suggestion about the estimation method are appreciated.Or does anyone know papers about estimation of jump parameters ?

I don't know what a Merton jump model is. But I would assume that jump parameters are evn more stochastic than stochastic volatility. So an immediate cross-asset analysis might be as useful as a long historical analysis of a single asset.The more temporary jump parameters are, and the more fleeting jumps are as their rate increases and the volatility of their size goes up, the more different sets of parameters could explain the same observation. And the more different observations you could get from the same parameters.So I would feel more comfortable with the jump parameters on stock or currency C, if I had also just come up with some jump parameters for similar stocks or currencies A and B. And this could easily be tested, by seeing whether outlier best fits hold over neighboring intervals, or whether they revert towards the best fits measured in similar things on the same period where the outlier was calibrated.