Can someone explain 'ex-post risk can be a biased measure of ex-ante risk'.

[chovies81]

This is on pg 18/19 of volume 3 (CFA text). It makes no sense. Specifically the concept that looking backwards we underestimate ex-ante risk and overestimate ex ante anticpated returns. The text states ” Suppose that bond prices anticipate a small chance of a central bank policy change that would be very negative for inflation and bond returns. When investors become aware that the risk has passed , bond prices should show strong gains. Ex post bond returns are high although ex ante they were lower. Because the bank policy change did not occur, it may be overlooked as a risk that was faced by bond investors at the time. An analyst reviewing the record might conclude that bonds earn high returns in excess of short term interest rates.
After typing this I am starting to wonder whether we underestimate the risk because we overlook the risks that may have occured at the time? If we overlook the risks why would we overestimate retuns? Seems ambiguous….

 

[malnoll]

It is specifically discussing the biases of looking at historical returns and policy changing events. If you simply looked at historic returns, you would just see a number for ex post. Understanding why a specific year’s return was overly high/low allows a more relevant number to use. Hard for me to explain, but the main point is understanding the numbers, beyond just looking at a number. Understand why that number was what it was.
Your point above about underestimating historic risk, because the policy risk at the time is overlooked, is right. But, if you underestimate risk, you underestimate returns, and when the policy event does not happen, the run up in prices is not seen by the analyst as an effect of the policy event not happening.
Think about the underestimation of risk as a lower required return or discount rate. It would be lower, and you expected return would be incorrectly lower as well.
Hope this helps. From what I hear, L3 is all conceptual for the most part.

 

[chovies81]

Thanks this is conceptual. It does make more sense looking at it from your vantage.

 

[Chi Paul]

I think of it as being concerned about a potential (negative) risk, that potential risk never happens, so therefore, looking back historically that risk that never happened is not captured by a standard deviation.