That certainly seems true, but how much of that is because by doing so you get to go to a school like Caltech, Harvard, or MIT? A more interesting question to me would be whether American IMO contestants have a better chance of doing good mathematical research than other math majors at their school. But it's hard to come to any definite conclusion, especially given the small sample size and the difficulty of classifying what is "good research."