The reality is that estimating underground rail construction costs is significantly harder and riskier than above ground [1]. According to the AACE's classification system for cost estimates, the current number is a borderline class 2/3 estimate which means the estimate could be off as little as 5% or as much as 600%. Historically tunneling projects have ended up in the upper half of that range.
[1] http://web.mst.edu/~rogersda/umrcourses/ge441/Too_Dangerous_...
The point of citing both was definitely to show that you can have a pretty large variance in costs, even if you're the world expert in building these things in earthquake-prone areas.
Chinese HSR had a fatal one within three years of the line opening: https://en.wikipedia.org/wiki/Wenzhou_train_collision