I encountered this example 31-17 in a book that I am reading, and the author says that the Neapolitan chord follows the "falling fifth pattern".
I guess that would be the case if the Neapolitan chord is interpreted as a derivation of ii full diminished chord. But the bass tone is still 4th scale degree.
So when falling fifth is in discussion, does a person follow what is in the bass or the chord label?
In the particular case of your example, and generally for any harmony example that is given with Roman numeral root analysis, the interval movement is referring to chord root movement.
However, if you are working with figured bass harmony, it if often described in terms of bass movement (bass not root) and interval patterns will refer to the bass movement. For example, "bass moving sequentially by down a fifth, up a fourth."
An easy way to make the distinction is to pair up root analysis with root movement, and figured bass with bass movement.
Also, generally, if the discussion is about voice leading or relative motion, intervals of motion will be about the actual voice movements and not about root progressions. Usually, that is obvious, ex. "bass ascends by P4 to tonic, soprano ascends by m2 to tonic"; clearly not describing root movements. But there may be cases not so explicitly stated where you can assume voice movement descriptions are not about roots.
I generally agree with Michael Curtis's answer: most (though not all) modern harmony textbooks tend to assume "chord root" motion when speaking of a "falling fifths" sequence. When speaking about figured bass patterns, however, it's more typical to reference an actual bass line.
So, a progression like I-IV-viio-iii-vi-ii-V-I will often be regarded as "descending fifths" in many textbooks, regardless of what inversion the chords may be in.
I would, however, note a detail about the specific example offered in the question. It's very typical for such a sequential pattern to break out of the literal fifths in the bass to move toward a cadence like this, as it produces a stereotypical 4-5-1 (that is, fa-sol-do) bass pattern. This particular three-note motion dates back many centuries as a standard cadential bass pattern, and it's much more common than a 2-5-1 (re-sol-do) pattern.
Thus, you'll see "descending fifths sequences" almost always break the bass pattern near a cadence to do this if resolving to tonic. This pattern is even more critical for Neapolitan chords, which (as the question notes) are a variant of a supertonic chord. But having the bass continue in the fifth pattern would result in a tritone leap (from A-flat to D-natural). Classical counterpoint would generally frown on such leaps unless they are "resolved" melodically.
This is one reason the Neapolitan chord itself is often specifically called the "Neapolitan-sixth" and almost always appears in "first inversion." It's not really a functional "flat-II" chord in older styles, instead just being a predominant harmony that's chromatically altered.
In sum, with this particular chord, you'll almost always see a break in any descending fifths pattern anyway. But the broader context of moving toward a cadence, and making a 4-5-1 bass line, is the general underlying assumption determining the bass line here.
You can assume that the author did not make a mistake and infer that the author is considering the N6 as a ♭II. If you consider the bass note then there is no falling fifth and the author's statement makes no sense.
So when falling fifth is in discussion, does a person follow what is in the bass or the chord label?
Both of them, supposing the chord label is derived after the harmony analysis of the score (including, but not limiting to, the bass path).
Almost in all the cases, the Neapolitan chord is presented in the first facing, so the 
would correspond to the 
and, since the II♭ of G minor is A♭ it actually "virtually" follows the falling fifth pattern (G - C - F - B♭ - E♭ - A♭).
