It is easy to budget a project that is something you have done before. It is much harder to do so for something new. Since nobody has built a James Webb telescope or a MSL before it is very difficult to budget properly. The only way you can do it is to do a best estimate and then add a substantial margin for uncertainty - 40%, 50%, whatever.

But acountants, especially government accountants, don't like large margins. They don't like large amounts of cash being tied in reserve. It is easier to get aditional funding as needed than get a realistic margin.

In such a system all budgets for advanced projects will tend to over. If it is anybody's fault, it is the fault of the accounting system. It is always easier to blame people and orgsnisations that an accounting system.

The only other option is to stop doing advanced projects. Each space mission will be just an incremental advance on the previous one. Over time, however, the improvement will be marked -compare Mariner IV and the Viking orbiters. But NASA has not done that sort of series missions since the 70s.

It would also be interesting to see how much earmarks inflate costs too.