An analyst is estimating whether a funds excess return for a quarter is related to interest rates and last quarters excess return. The regression equation is found to have unconditional heteroskedasticity and serial correlation. Which of the following will be TRUE? Parameter estimates will be: A) accurate but statistical inference about the parameters will not be valid. B) accurate and statistical inference about the parameters will be valid. C) inaccurate and statistical inference about the parameters will not be valid. D) inaccurate but statistical inference about the parameters will be valid.

Both effect the standard errors but not the coefficients. A. I have a feeling I am missing something due to the lagged variable.

ye the lagged variable damn isnt AR based on a lagged dependent var as a independent one… then why is the model misspecified … i m doin the advanced Qns of quant from Q bank damn they r toooo much !!

Is it A? It would be misspecified because this isn’t an AR model.

no the answer is C

Damn, that was the other choice I was thinking of because of the lag.

UCH is no big issue. Serial Correlation will cause wrong inferences and inaccurate estimates C? EDIT: Sorry, the answer was already given away…

yes

^^^Nice catch, I just read heteroskedasticity and didn’t read closely enough. Dinesh: When they say parameter estimates, do they mean just the variables or the coefficients?

I think they are talking about coefficients here, no?

yes co-efficients

Yes, they are talking about coefficients here. Term “forecast” is generally used for independent variables.

I always thought that the estimates of the coeffiecents were not affected by serial correlation, just the standard errors… Man I suc at quant

It has to do with the lagged independent variable I think.

LanceTX Wrote: ------------------------------------------------------- > I always thought that the estimates of the > coeffiecents were not affected by serial > correlation, just the standard errors… Man I suc > at quant If you don’t have lagged term in regression equation yes only standard errors get affected. If you have lagged term in regression equation both get affected. Refer to Model Misspecification portion for this.

It makes perfect sense now. At least I knew something was wrong haha.

Thanks Kabhii!!

Another Q Bank Question regarding Ethics An analyst finds a stock with historical returns that are not correlated with interest rate changes. The analyst writes a report for his clients that have large allocations in fixed-income instruments and emphasizes the observed lack of correlation. The clients with allocations of fixed income instruments are the only clients to see the report. According to Standard V(B), Communication with Clients and Prospective Clients, the analyst has: A) violated the Standard concerning fair dealings with all clients. B) violated the Standard by emphasizing the information concerning the correlation. C) not violated the Standard. D) violated the article in the Standard concerning facts and opinions.

C or D . . . I will go C.

A