Higher correlations during crises or not?

We are presented with the argument that during times of crises, foreign markets are typically more correlated. We are also presented with the argument that this is a statistical aberration. Which one is correct or could both be used as a defense of an answer? I would think that the argument against higher correlations would trump the argument for them, but who knows what CFAI wants to see on the actual exam. I’m worried that a MC question will have a statement made by someone to the effect of “Financial markets are more correlated during times of crises” and you have to state if this is correct or not.

Correlations are higher as a result of higher volatility and not as a change in a relationship itself. It is a result of a way correlations are calculated (in statistical sense)

So are correlations “actually” higher? I thought the text says that they really aren’t. This is the type of question where you can easily overthink it and get it wrong on the exam.

i think text is arguing that “real” (if I can make up that term) correlations or true relationships between two variables, as csk said, are unchanged or changed slightly, but correlations we calculate are higher due to increased volatility in times of crisis

I cant really answer the question of why, Joey would be a better person to ask

I am not a quant, but I think that correlation goes up if volatility goes up. This means that correlation is not “100% true”… Ok, but that is only mathematics and statistics, in times of crises both emerging mkts and developed mkts go down. I think these are 2 separated things.

It looks like correlations are higher during times of crises, but it is due to higher volatility as opposed to higher co-movement. I guess either way, the correlations are higher so diversification benefits decrease.

there is co-movement, that is why diversification benefits decrease if you measure taht co-movement with correlation, that is the problem. the increase in correlation comes not only from co-movement, but for any increase in volatility

That’s interesting. Is the CFA book making the case that correlations are the wrong way to look at co-integrated processes? If so, I might agree with that. It’s kind of strange to say “Those are the wrong measure and they are too high”. I can tell you that the usual paper cited with this result is Boyer, Gibson and Loretan (1999) which is a correct paper but irrelevant to capital markets. Most of the papers that cite this, e.g. http://www.bis.org/publ/confer08k.pdf are simply wrong.