There is evidence of heteroskedasticity, serial correlation, and multicollinearity. Viewed seperately what effect will this have on the T-Stat and type of errors for each. GO!
Hetero and Serial - Tstat too high …Type 1 error Multi- T stat too low, High R2, Type 2 error
Hetero= SE too small, t-test too big, Type I error Serial= SE too small, t-test too big, Type I error (positive…opposite for negative) Multi= SE too big, t-tests to small, Type II eroor
nice mw, nice
Heteroskedasticty = smaller standard errors - type 1 - robust standard errors Positive serial correlation = smaller standard errors - type 1 - hansen/robust standard errors Negative serial correlation = larger standard errors - type 2 - hansen Multicollinearity = biased coefficients and larger standard errors - type 2 (not so sure on this one) - drop variable
how do you remember this sh!t… ??? if you have any acronyms please help… http://www.analystforum.com/phorums/read.php?12,756309