Clyde Schechter All have to go back to the main proposition:
If analyst i follows firms j, k, and l in quarter t, the fe_other on analyst i following firm j in t this observation is calculated as other analysts’ average forecast error on firms k and l (not j the focal firm) in quarter t‒1.
For example if I am the analyst who follows Apple (j), Netflix (k) and Facebook (l) in 2022q3 (t), the fe_other on - me the analyst, Apple (j), 2022q3 (t) - this observation would be calculated by using the average forecast error of other analysts on Netflix (k) and Facebook (l) in 2022q2 (t-1).
It doesn't matter whether I followed Netflix (k) and Facebook (l) in preceding quarter, 2022q2, as I don't need my own forecast error on Netflix (k) and Facebook (l) in preceding quarter to calculate the fe_other on Apple (j) in current quarter, what I need is other analysts' forecast errors on Netflix (k) and Facebook (l) in preceding quarter.
That's why it is important that a given analyst follows more than one company in quarter t. If this is not the case, we could not even proceed to consider anything about other analysts and preceding quarter.
I'm referring to 2021q2.
The thing is, this analyst does not follow any other firm in 2021q3 (t). Say if this analyst is following Apple in current quarter (t), but he/she does not follow Netflix or Facebook in current quarter - no other firms at all. In this case, we even don't need to proceed further to consider anything about other analysts and preceding quarter. fe_other for this analyst following Apple in t has to be a missing value.
I hope that up to this point this question could be solved. E.g. This analyst #53035 follows both Apple and Netflix in 2021q3. fe_other on this observation - analyst #53035, Apple, 2021q3 - would be the average forecast error of other analysts (not #53035) on Netflix in 2021q2. Same logic holds if now our focus is on - analyst #53035, Netflix, 2021q3 - The fe_other for this observation would be the average forecast error of other analysts on Apple in 2021q2.
If analyst i follows firms j, k, and l in quarter t, the fe_other on analyst i following firm j in t this observation is calculated as other analysts’ average forecast error on firms k and l (not j the focal firm) in quarter t‒1.
For example if I am the analyst who follows Apple (j), Netflix (k) and Facebook (l) in 2022q3 (t), the fe_other on - me the analyst, Apple (j), 2022q3 (t) - this observation would be calculated by using the average forecast error of other analysts on Netflix (k) and Facebook (l) in 2022q2 (t-1).
It doesn't matter whether I followed Netflix (k) and Facebook (l) in preceding quarter, 2022q2, as I don't need my own forecast error on Netflix (k) and Facebook (l) in preceding quarter to calculate the fe_other on Apple (j) in current quarter, what I need is other analysts' forecast errors on Netflix (k) and Facebook (l) in preceding quarter.
That's why it is important that a given analyst follows more than one company in quarter t. If this is not the case, we could not even proceed to consider anything about other analysts and preceding quarter.
For the first row - analys #70 who follows company #1174 in 2021q2. Apparently, this analyst does not follow other company in the same quarter. So there is no peer average forecast error (in t-1) on other firm that this analyst follows in current quarter (as he only follows one firm in current quarter!). fe_other for this observation will therefore be a missing value (i.e. '' . '')
When you say this analyst does not follow other companies in the same quarter, which quarter are you referring to? 2021q2? 2021q1 (which would be t-1)?
When you say this analyst does not follow other companies in the same quarter, which quarter are you referring to? 2021q2? 2021q1 (which would be t-1)?
For the second row - analys #70 who follows company #512 in 2021q3. Again, this analyst only follows one firm this quarter. fe_other will be a missing value.
Yes, but in t-1, q2021q2, he/she does follow another firm, namely 1174. So why isn't the result the value of fe_own in 2021q2, namely -.0004757?
Yes, but in t-1, q2021q2, he/she does follow another firm, namely 1174. So why isn't the result the value of fe_own in 2021q2, namely -.0004757?
Finally when we move to row 23 - analys #53035 who follows company #427 in 2021q3. We could see that this analyst also follows another company #512 in the same quarter. Great, that's what we love to see. Now, we have to figure out peer analysts' (aka analysts other than #53035) average forecast error on company #512 in 2021q2, i.e. previous quarter. With my own manual calculation, the fe_other for this would be 0.003795.
Apparently it is important that this analyst follows more than one company in 2021q3. But I don't understand why that is relevant for that observation: what I thought should matter is which companies the analyst followed in t-1, i.e., in 2021q2.
Apparently it is important that this analyst follows more than one company in 2021q3. But I don't understand why that is relevant for that observation: what I thought should matter is which companies the analyst followed in t-1, i.e., in 2021q2.
Comment