Data geek niggle - it's pretty deceptive to start the graph at $200 million instead of zero. It makes the drop in 2012 look larger than it really is. Granted, the effect isn't huge in this case, but its still very bad practice.