I'll start by setting the tangential part of the question aside: In quantum field theory there are currents and charges which are conserved classically, but not quantum mechanically, that is, there are violations of these conservation laws to order $\hbar^2$, and hence the symmetries hold classically but not quantum mechanically. (See below for what that means.) The non-conservation pieces are called [**quantum anomalies**][1] and the corresponding symmetries **anomalous**. (Other symmetry breaking phenomena require subtleties of QFT:Goldstone's theorem.) 

Now for your core question: Classical mechanics is, "basically" an effective theory of the full theory of the world, quantum mechanics, that is, its [ Classical limit][2]. 

Mathematically, that means that all quantities of interest are functions of the dimensionful $\hbar$, so, normally they entail a classical piece without $\hbar$ and a "quantum correction piece" involving powers of $\hbar$ normalized by characteristic actions, angular momenta, etc... so as to be dimensionless. When *these* quantities assume macroscopic values, (so, for engineering angular momenta or actions), these powers, and so quantum corrections are *infinitesimally small*, and thus ignorable. The leading term in this notional expansion is called **the classical limit**, and many texts on the quantum oscillator remind the student how small such correction effects beyond that limit are, in practice, and when they step to the fore and control atoms and crystals, etc. 

A fascinating technical feature happens in this limit---actually dramatized in the struggle of humanity to go beyond this limit, in the 20th century: The mathematical rules describing the full QM theory are very different than the math of the  classical limit. QM was worked out in the 1920s, mostly, with a resolute insistence on keeping away from the confusing classical limit, and violating its mathematical structure with giddy revolutionary abandon!

The reformulation of the QM formulation in a way that made that limit plausible (beyond the early and reassuring Ehrenfest theorem) was *only* 69 years ago: the [Phase space Formulation of QM][3]. In that formulation, it is manifest at every step how QM is a "deformation" of classical mechanics (a mathematese and engineering term: basically means systematic theory of correction) and how, conversely, classical mechanics is a simplified summary effective description of QM. (For example, Dirac, in 1933, Ref 1, explained how the principle of least action follows from the $\hbar/S\rightarrow 0$ limit of amplitudes through constructive interference.) Observables are now described by $\hbar$-dependent Wigner transforms of operators, whose detailed behavior "morphs" that of classical observables in phase space.

Your question is really a question on deformation quantization, but without conscious reference to it!

References 

1 Paul A. M. Dirac, "The Lagrangian in Quantum Mechanics", *Physikalische Zeitschrift der Sowjetunion*, **3** (1933) 64–72}} 

2 Also see J H Van Vleck, ["The correspondence principle in the statistical interpretation of quantum mechanics" ][4], *Proceedings of the National Academy of Sciences of the United States of America* **14** (1928) 178–188, (doi: 10.1073/pnas.14.2.178)  


  [1]: http://physics.stackexchange.com/questions/tagged/anomalies
  [2]: https://en.wikipedia.org/wiki/Classical_limit
  [3]: https://en.wikipedia.org/wiki/Phase_space_formulation
  [4]: http://www.pnas.org/content/14/2/178