This is a very ambitious idea.  There are several implementation problems, however, that would need to be addressed.  First, simply having a good database of ALL laws (even just from the US) is a very large project.  Although there are commercial groups such as West and Lexis (or whatever they have been renamed) that are engaged in this task, their systems are highly proprietary and, in my opinion, not suited for algorithmic research.  Second, moving from the text of the statute to an economic analysis of beneficiaries and victims is incredibly difficult to do algorithmically.  Moreover, the unit of analysis is not clear: one chunk of a statute may help, say, a corporation, while another linked in chunk may hurt it.  

There's a conference this coming week at the University of San Diego (http://www.sandiego.edu/law/school/events/detail.php?_focus=46107) in which academics interested in this field are showing beginning forays, but what you are describing is, I think, a long, long way off.