The only reason I can think of to include a outright recommendation the authors can see is if the "feel" of my comments and the decision don't necessarily align naturally. For example, if I've issued a lot of criticisms, but most of those are "With fine tuning, this would be an outstanding paper", or hoping to see what is an adequate treatment of really spectacular data turn into a great treatment, I might not that despite the page of suggestions, it could probably fly as is.
At the same time, I've reviewed a paper with a very small number of comments, but in those comments have been tempted to use words like 'fatally flawed'. So including a 'I would reject this paper' comment might help with "There are only three things you need to change, but you need to take them seriously."