in the CFAI text, Reading 13, Page 366 they go through a sample question evaluating the significance of some autocorrelation values. They give the following: Lag-----AutoCorr--------Std Error--------t-stat 1---------0.0677----------0.1302---------0.5197 2---------(-0.1929)-------0.1302--------(-1.4814) 3---------0.0541----------0.1302--------0.4152 4---------(-0.1498)-------0.1302--------(-1.1507) 59 Observations and 2 parameters. Using a 5% significance level which of the autocorrelations are significantly different from 0? They state “Table 4 shows that none of the first four autocorrelations has a t-statistic larger than 1.50 in absolute value.” Where does the 1.50 value come from? I am loosing my mind and ready to move onto my econ review already.

That is the largest t-stat generated from the test of autocorrelation (in absolute terms). The actual value is -1.48. They are rouding to 1.5. This number is compared to your t-critical value of 2.0. Since all are less, none are significant. Confusing the way they worded it.

I think it should read 2, not 1.5.

No. That is what I thought at first too. Edit: I am not trying to sound like a jerk here.

the book actually just want to make a point that the t statistic is very low and unlikely to be significant. if you look up the t value for the .05 significance level and df = 57 (two tails) t_0.05_95% = 2.0. So none of the lagged terms demonstrate the significant correlation because their t statistics are below 1.5, let along 2.0. actually you don’t need to look up the t table because you know the z_0.05_95% = 1.98 and t_0.05_95% > z_0.05_95%.

wanderingcfa Wrote: ------------------------------------------------------- > in the CFAI text, Reading 13, Page 366 they go > through a sample question evaluating the > significance of some autocorrelation values. > > > They give the following: > > Lag-----AutoCorr--------Std Error--------t-stat > 1---------0.0677----------0.1302---------0.5197 > 2---------(-0.1929)-------0.1302--------(-1.4814) !!!This is your largest t-stat. In absolute terms in is about 1.5. > 3---------0.0541----------0.1302--------0.4152 > 4---------(-0.1498)-------0.1302--------(-1.1507) > > 59 Observations and 2 parameters. Using a 5% > significance level which of the autocorrelations > are significantly different from 0? > > They state “Table 4 shows that none of the first > four autocorrelations has a t-statistic larger > than 1.50 in absolute value.” > > Where does the 1.50 value come from? I am loosing > my mind and ready to move onto my econ review > already. See above.

Ok, good. I keep thinking that I should be comparing it to the critical value of 2 but when they said “1.5” I could not figure out why they said 1.5. Thanks everyone (especially mwvt9 since he was first)…haha