There is considerable disagreement on this subject. The arguments against 14bit recording are
1. The additional bit depth does not carry useful information, except in extreme cases (like adjusting the exposure by five or more stops in raw processing). The shades between the 12- and 14bit values represent noise only.
2. The additional information can not be utilized due to the imitations of 8bit JPEG and of today's printers and monitors.
Point 1 can be discussed endlessly.
Point 2 is faulty, for
a. the current 8bit JPEG is on life support. It will exist for many decades to come because of the billions of existing images, but its active life is virtually at its end,
b. printers are already avaliable with more than 8bit support. Current high-resolution monitors have contrast ratios like 300:1 to 1000:1. Lower resolutions, like HDTV go higher already. New technologies are already available, though too expensive at the moment, with contrast ratio 10000:1 or more. These monitors will require more levels of shades, i.e. greater bit depth.
Storage should not be a big consideration, CF cards are cheap, hard disk space is dirt cheap. If the speed is a concern, then go 12bit, compressed, but losslessly. There is no excuse for lossy compression.