In support of Johanna Rubba's excellent clarifying statement on "BE supposed
to" let me make two comments:
 
[1] The OED gives, among its many historical meanings of "suppose," these two:
"to expect" and "to assume." I think it would make sense to suppose (forgive
me) that originally the expression "You are supposed to..." was no different
from "You are expected to...." In other words, a passive construction. Not all
passives have identifiable agents. We use the passive, for example, to indicate
what people (or experts) generally do: "A copper pan is used when beating
eggs."
 
[2] The evolution of "BE supposed to" into a model (something that many seem to
feel instinctively is happening, or has happened) can be discussed in another
way (besides the historical perspective Johanna introduced). We can talk about
word classes as defined by  "prototypical" cases--examples that fit the
category perfectly. Then we can refer to instances that fit the category
imperfectly as non-prototypical cases. Thus, "old" is a prototypical adjective
because, besides taking certain suffixs (-ness, -er), it can fit in either of
these two syntactical positions:
 
The _____ man seems very _____.
 
"Old" has all the features of an adjective. But neither "alone" nor "main" are
prototypical adjectives because they can only fit into one or the other of
those slots: "His main enemy is very alone," but not "His alone enemy is very
main." And "main" doesn't take "-ness" gracefully.
 
Our discussion, then, would focus not on finding the right label, but on noting
which features a construction posseses and which it lacks in relation to
prototypical cases. And we can note that a non-prototypical part of speech may
share characteristics of more than one part of speech.
        --Bill Murdick