"replacement-level players should be valued at no more than the league minimum."
That seems like a pretty fair statement, doesn't it? It's a tautology - if a given player can be replaced by freely-available talent, then there's no reason to pay him more than anyone else.
But for some reason, economist J.C. Bradbury disputes that notion in his new book, Hot Stove Economics. You can read some of analysis of why this is wrong in baseball over at the book blog. And here's what happens when you make this assertion and try to apply it to hockey - you get completely implausible results:
Mike Brodeur = +0.73 Wins
Martin Brodeur = + 0.62 Wins
This was through the 2010 trade deadline. For the full season, Martin is at 1.51 Wins in this replacement-free method. Let's break up the Brodeurs' seasons and see what we get:
Mike Brodeur = +0.73 Wins in three games, 12/19, 1/14, 1/16 -> 87 shots, 84 saves
Part 1: Martin Brodeur = +0.73 Wins in three games, 11/7, 12/21, 1/27 -> 87 shots, 84 saves
Part 2: Martin Brodeur = +0.77 Wins in 74 games -> 1917 shots, 1752 saves = .914 save percentage
Now, is it possible that Martin Brodeur, whose record was 43-25-5 in those 74 games in the second set, was worth the same amount to his team? Let me cherry-pick six games and let's look again:
Brodeur = +1.51 Wins in six games -> 174 shots, 168 saves (11/12, 1/5, 1/12 additional games)
Brodeur = 0 Wins in 71 games -> 1830 shots, 1668 saves = .911 save percentage
So I took six reasonably good games by Martin Brodeur which were basically equal to his entire contribution over the course of the season, and the remaining 71 games - where he was a league-average goaltender - were worth nothing to the New Jersey Devils? In 2005-06, Martin Brodeur played 73 games for New Jersey and posted a .911 save percentage. Was he worthless? Unless you accept the concept of replacement level, you will always get this wrong.