The market for NHL players is a fairly efficient one - unlike, say, baseball in the 1980s, where a little spreadsheet analysis would tell you that on-base-percentage was a low-cost path to success. So aggregate league-wide salary allocation is actually a pretty good reflection of the true value of talent by position:
| Marginal$ | F | D | G |
| 2010-11 | 757.8 | 402.6 | 101.4 |
| 2009-10 | 679.6 | 357.6 | 98.3 |
| 2008-09 | 657.0 | 397.8 | 97.3 |
| 2007-08 | 574.9 | 299.5 | 91.5 |
Those are marginal dollars - actual dollars minus the league minimum salary per player. Converting to percentages, we get:
| F | D | G | |
| 2010-11 | 60.1 | 31.9 | 8.0 |
| 2009-10 | 59.8 | 31.5 | 8.7 |
| 2008-09 | 57.0 | 34.5 | 8.4 |
| 2007-08 | 59.5 | 31.0 | 9.5 |
| AVG | 59.1 | 32.2 | 8.7 |
So that means - according to GMs - forwards are individually worth 19.7% of total value, defensemen 16.1% and goalies 8.7%. After years of paying big salaries for league-average performance, teams have become smarter about goaltender acquisitions, and the salary allocated to goalies is dropping rapidly.
Player valuation in the NHL is hardly perfect, but this payroll allocation is reasonably close to optimal. If you think these splits should be significantly different, you'd better have some very good evidence!