Alan Blinder and Alan Kreuger's surveyAlternative Measures of Offshorability states the following:

Jobs that can be broken down into simple, routinizable tasks are easier to offshore than jobs requiring complex thinking, judgment, and human interaction. [Blinder/Krueger; 2009, 7]

Offshorability appears to be particularly prevalent in production work and in office and administrative jobs. By industry group, it is most common in manufacturing, finance and insurance, information services, and professional and technical services. More educated workers appear to hold somewhat more offshorable jobs. [Blinder/Krueger; 2009, 39]

So this begs the question, which part of my definition (see another post) of structural engineering is offshorable?  I would venture to say it is the applied science, mathematics, and code verification portion (about 20% of what we do).  Anything that can be done automatically from a computer is offshorable (analysis models, drafting, science, etc).  Everything else (design) is not.  That being said, I do not think analysis and drafting would be done well overseas without having an engineer or drafter within the locality of the project review the work.  The local experience is invaluable in our field.  For example, the outsourcing of steel drafting (called detailing) has become a major problem in the steel fabrication business.   They charge very little overseas (detailers overseas charge $5-$30/ton as opposed to $60-$80/ton locally) but there are so many mistakes that can arise by using overseas detailers.  The 20k savings of using an overseas detailer may end up costing the fabricator 40k in extra welding costs or mistakes in interpreting the AISC code.

Think about what a robot can do - that part of engineering is offfshorable.  Engineering is a complex task that involves high levels of judgment and experience.  It also involves human interactions (meeting with architects, site visits, working in collaborative teams, etc).   None of this is offshorable.  Our scientists and mathematicians may need to worry more about this issue than we do.  The applied science or code procedure portion of what we do may be considered offshorable, is only 20% maximum of what we do.   Our software will continue to do more and more code verification and analysis, but again that is still a small portion of our daily activities as an engineer.

As an example of why even the applied science portion cannot easily be done by a computer, see my article Twilight of the Idols.  It compares a terrible floor system solution from a computer to a good floor system solution from an experienced engineer.