Most of the talk here has to do with the pure physics argument, but there is more at play here. The weather indoors is always perfect with no wind and typically good temperature with no precipitation. With mondo 200m indoor tracks that are built up on trusses (faster than hydraulic tracks due to the extra bounce) the energy losses from the turns are pretty damn small. Overall I think all of the conditions from each scenario almost even out, with outdoors being a little faster but the margin is much closer than the record books suggest.
For what it's worth, my best years I ran virtually identical times indoor (4:08.27) and outdoors (4:08.50) for the mile. The next year I ran 4:06.50 indoors which was marginally better than my outdoor 1500 pr (3:49.50) on conversion.
I am of the opinion that the only reason indoor records are noticeably slower than outdoor records, especially in middle distance, are because the money is on the track in the summer, not in the winter.