How is df calculated in Zellner's g-prior in Bayesian regression?

[Red Owl]

I would appreciate advice on how to calculate df for Zellner's g-prior in Stata's Bayesian regression.

Consider the Zellner's g-prior specification in the Bayesian regression in Example 6, p. 22, at http://www.stata.com/manuals/bayesbayes.pdf and as described at http://www.stata.com/manuals/bayesba...f#bayesbayesmh .

sysuse auto, clear
bayes, prior({price:}, zellnersg0(3, 50, {sigma2})) burnin(5000): regress price length i.foreign

In this example, n = 74, dimensions = 3, df = 50.
The levels of length = 47 and the levels of foreign = 2.

Thanks for any help you can provide.

Red Owl
Stata/IC version 15.0

Edited to correct p. number of Example 6.