It is well-known that asset return correlations of stocks increase during market downturns. But are there any general properties derived from empirical observation or evidence regarding by how much magnitudes in correlations change before and after the downturn? e.g. based on either the levels of correlations, or individual return distribution properties per pair, before the downturn.
For example, before the market collapse, the pair-wise correlations between 3 hypothetical stocks $A$, $B$ and $C$ are
corr(A,B) 0.1 corr(A,C) 0.3 corr(B,C) 0.6
After the downturn (or maybe sometime during it), what can be said about how much these correlations increase (given that we know they will not decrease).
Could it be that each of the three correlations double? increase by 10%? Only one of the three correlations increases by 10% while the other two only increase by 50%? If so, which.
Do low (higher) correlations before, signal higher (much higher) correlations after? Empirically, what has been observed? are asset expected returns, distributions or variances before the downturn of any use as indicators of the impending bump in correlation?