LoL Totally undecided position after 11th move. I launched Stockfish and it gives tons of more or less equal possibilities including your 12. h3 as well as 12. Kb1 and 12. Rhe1 evaluated at about -0.2.
Are U sure that saving tempi is such a primordial task during the middlegame? 
When you play opposite castling positions , tempi is the most important thing. A faster consolidation against your opponents King can give a lot of practical chances.
Engine evaluation in this position it's not something you can take it seriously. I don't believe engines understand this position. My feeling playing this game after 12.h3 was that White is much more worst than -0.2
Find 7Nd2 quite an unnatural move, even though quite a few strong players have played it. It's designed against h6, g5 and ne4, but I think if white fears that after 7Qc2, or 7e3 can play 7...h6 8BxN.
Don't think white has much chance of a opening advantage after 110.0.0. Bb4 can be useful, whilst doesn't seem much value to nd2. Think more point to it in positions with h6, Bh4, although even there black should be ok.
7.Nd2 it's intriguing inviting Bxc3 and also keeps the c pawn on c6 for quite some time when Black in the Ragozin doesn't want to play an early h6-g5 waiting White to commit short castle. In general the exchange variation of the Ragozin it's famous for the unbalanced sharp middlegame but Whites "playing for an advantage " is hidden on taking risks not let's play conservative and we will see, otherwise we talk for one more sharp theoretical draw.
The idea to castle 0-0-0 is to deny the usual game and risk your King safety for a uncertain attack. Both Bb4 and Nd2 are lost tempi waiting to find their role. The point is...can White accelerate he's attacking conditions or not and how?
It seems GM's don't believe in it...and having see a lot of opposite castles in QGD games, I don't blame them.
Pert in similar position with h6, Bh4 included doesn't consider 120.0.0. Guess doesn't consider it dangerous, but his main continuation is 8...Be7,instead of 8...c6, intending quick c5. This probably shows up the bad side of Nd2 even more.
On the plus side queen-side castling gets fresh position. Don't think there is any known theory about it.
After 8...Be7 Black is quite passive when the quick c5 doesn't mean Black likes to open the c file and White knows about it. Finally we end up pushing e4 where Nd2 shows it useful nature to support a better piece placement, unless your name is GM Matlakov and you enjoy wild plans in style Qf3-g4!? e.t.c
How theoretical value have all this it's not important in front of the practical chances you get. In the end the better player on the board wins.
Now I remembered Carlsen losing an opposite castling game in a QGD with Bf4 against Ivanchuk back in 2010 or 11 . Didn't he knew the risk was taken in a theoretical uncharted territory ? He definitely knew but in such positions you assume that your opponent is gonna miss something...
In comparison with club players like us who don't face Chuky every day, I guess it worth the risk...and it's a great fun after all, to play such position.
11.0-0-0 is interesting, although it has been played only once in a hi-level game.
The problem with engine evaluations in very double-edged positions of a strategical nature (when the final assault can materialize 20 or 30 moves after the evaluated position - or even much later, in case of transition to an endgame) - is twofold.
Firstly, the engine often changes its mind once you SHOW it a resource that it failed to appreciate
(the engine gives +0.2 for a couple of moves, and then you feed a line into it, that it formerly evaluated as +0.00. You play several moves into the line - and suddenly the engine sees it as +0.78 and updates its evaluation backwards).
Secondly - there are positions that a computer will defend but a human being will not - and this is where the engine eval will really lead a player astray - if in studying that position with the engine, that player doesn't understand how difficult that position will be to defend in practice.
I had a game as White before couple of months in a rare line to the Ragozin defense. I know that this positions meant to be doubled edged and its difficult to evaluate them, perhaps this is the reason why GM Vladimir Barsky doesnt mention anything about it ( The Ragozin complex)...but still my intuitive thinking cant stop lurking for improvements.
In practice i have found only one game in top level by GM Sakaev back in 2007 which went totally different from mine , so i was wandering if someone have samples from correnspondence data that either confirm my instict the line can be improved for White or at least has personal experience and analysis.