arima does not return probability value in e(p)

[dwcarley]

I am using arima in STATA 12.1. The display output looks fine, but the arima command always leaves the e(p) scalar empty (null, or "."); yet all other scalars, macros and matrices returned by arima are updated properly. This can be easily demonstrated by "ereturn list" after an arima command. "Prob > chi2" always shows a proper value between 0 and 1 in the display console, despite the e(p) scalar being empty. Has anyone else run into this? Any help appreciated.
Dave Carley

[Maarten Buis]

Here is an example in which I could replicate that behaviour.

webuse wpi1
arima wpi, arima(1,1,1)
di e(p)

We can recover those p-values using tricks discussed in: M.L. Buis (2012) "Stata tip 112: Where did my p-values go? (part 2)", The Stata Journal, 12(4), pp.759-760.

di chi2tail(e(df_m),e(chi2))

Note that the p-value is very small: 3.519e-68 = 0.000000000000000000000000000000000000000000000000 000000000000000000003519 (I may be one or two zeros off), which I guess is the reason for this problem.

[dwcarley]

Thanks Maarten, that took care of it.