The simple answer: Because mandatory harvest reports don't give the DWR any better information.
The long answer:
A few points by the article: http://www.mnr.gov.on.ca/MNR_E005349.pdf
1) Mandatory reporting does NOT ensure 100% reporting or 100% accuracy...in fact, sometimes hunters perceive an incentive to actually lie about their success or lack thereof--" Gamesmanship, or perceived dangers from giving true information may cause people to misrepresent the truth (i.e. lie), especially about killing an animal. This might be more prevalent for...hunters who may believe tag numbers would go down if kill rates are high or among individuals who might believe that reporting the harvest of an animal could lower their personal chance of getting a tag in future."
2) Costs--"When all costs are considered, mandatory reporting may cost about nine
times as much as an equal size voluntary sample and six times as much per report (see “The costs of big game harvest assessment”, page 9). Optimal sampling within the deer or moose programs would produce statistically valid harvest estimates with about 30 per cent of hunters sampled. In these cases mandatory reporting from all hunters could cost as much as 23 times more than optimal voluntary programs." Also, "Statistical methods operate on the principle that a representative sample can provide information which is as good as a complete count, but less (often much less) expensive. In many cases, a complete count is not possible. Mandatory reporting ignores the economic savings which optimal sampling could provide. It should be necessary to demonstrate that the economic efficiency of mandatory reporting exceeds that of optimal sampling (i.e. that a 300 per cent increase in costs produces at least a 300 per cent improvement in information value)."
3) Hunter surveys tend to overestimate harvest which errs on the side of conservation--"“Statistical Estimates” of harvest are not expected to be perfectly accurate. They depend on an assumption that the activities of hunters who report are similar to those who do not report or those who are not sampled. This is not always true. Voluntary sampling tends to overestimate harvest because hunters who feel they have something important to report (i.e. a harvested animal) tend to respond at a higher rate than those who do not harvest game. This error is on the side of conservation, but can be corrected, again by statistical techniques. Generally, statistically estimates may be either higher or lower than the true
harvest, but they are correct “on average”."
4) Mandatory harvest reports tend to underestimate harvest--"Mandatory reports would likely underestimate harvests. Few people would be expected to report killing an animal if they did not and more advantages might be gained by not reporting actual kills. Because harvest information is “added” for mandatory reports rather than “projected”(as with sampling), every animal which is not reported represents an underestimate of the true harvest."
5) A wealth of information aside from harvest is needed to best control big game populations--"Many factors influence the abundance of game and the allowable harvest.
These include habitat quality, productivity, predation, accidental mortality, as well as subsistence harvests and recreational hunting. Each of these differ geographically, probably annually, and they may interact in complex ways. The important point is that most of these factors are measured crudely or not at all. The value of obtaining extremely high quality harvest information at relatively high cost is undermined by having little or no information on other factors."
6) Mandatory harvest reporting rarely gives drastically different information than samples--"If all other things are equal and unbiased, a proper statistical interpretation of the harvest estimate for a hypothetical WMU would state that “the harvest was probably between 95 and 105 animals, and averaged 100 animals over the past three years.” Mandatory reporting would state “the harvest was 96 in year 1, 106 in year 2, and 98 in year 3.” There is no reason to believe that mandatory reporting would provide a totally different answer (like the harvest was 50 or 150 animals) and clearly the management decisions from both voluntary and mandatory assessments should be identical."