I noticed that some conference have different deadline for paper submission: an "abstract" submission deadline, before the usual "paper" submission deadline.
For example, on the International Semantic Web Conference 2013 webpage you can read:
Submission dates
Abstracts: May 1, 2013
Full Paper Submission: May 10, 2013
Why do they need the abstract before the paper? To estimate how many papers they'll get?
From what I have observed, having a specific deadline for abstracts is used for two main reasons: having a rough idea of the number of submissions and organise a bidding for the reviewers.
Having the number of submissions can help deciding of a possible deadline extension and possible to "recruit" more PC members or reviewers if the number largely exceeds the expectation.
Organising a bidding based on the abstract allow the PC members to indicate their preference for each paper (e.g., I want to review this paper, I could review this paper, I couldn't review this paper), so that when the actual papers arrive, the distribution is already organised.
In my field (chemistry/spectroscopy/chemometrics), the abstract decides whether you'll get an oral presentation or a poster (total rejection is extremely rare). The paper submission deadline is usually after the conference.
Once submitted, the paper undergoes normal peer-review for the journal it is submitted to, which doesn't have anything to do with the presentation at the conference. The only connection is that the conference organisers have spoken with the journal editors that they'll collect papers about topics presented at the conference in a special issue of the journal.
So the paper deadline is needed by the journal editors to make people submit in time so that the special issue will be ready at the specified date.
An unspoken reason is to enable conference presenters to prepare their abstract months before their paper or poster is ready, while they are still doing their research. Some large conferences, such as the American Geophysical Union, request abstracts months in advance of final submission. What's a struggling (or highly distinguished--both find themselves in the same predicament) researcher to do? Write something, and hope that by the time of the conference, the research meets or exceeds the statement of the abstract.
