Thanks to those who have replied so far
I agree with those who thinks some artificial methods (like bart or Gazzilli) would make this problem easier. On the other hand, the methods (or lack of methods) are used by several recent World Champions, and should not be unplayable (I have played with Gazzilli in this position in other partnerships).
I was a bit surprised that only one poster mentioned the form of scoring, for me this is the deciding factor. I would invite at IMP's (with 2NT) and settle for a partscore and hopefully a plus at Matchpoints (as mentioned in the OP this was what we played). This could also be a factor when deciding on partscore (spades or NT), it seems those posters who have written about possible denominations discusses mostly which game is most likely, not the chances of the same number of tricks in a NT or a spade partscore. This may also be an argument for settling for a partscore, since even if game is making some may go down in the wrong game, still giving us some MP's for staying low.
Partners actual hand is not so important, but in practice he held:
xx
Kxxx
A8xxx
xx
Not the worst hand he could have, but he would still probably pass 2NT/3sp or a 1NT opening. In practice there were only 7 tricks in NT and 8 or 9 (most defenders failed and gave declarer 9). 1NT= was worth exactly average. If I remember correctly dealers RHO had AT8x, QJT8, Kxx, Tx and LHO had 9x, 9xx, QJTx, A98x (playing in spades 9 trick comes in if declarer can ruff a club overruffed with an original trumptrick without letting RHO score a second clubruff/overruff).
John