Why? as a rough estimate from 2020 it is not that far off. The number of J-10s is IMO higher today as the number of J-16 and most likely J-20 too, but again for 2020 it is not that "deadly wrong" IMO?
Why do you think so?
I agree that it is the number of J-10s (specifically, likely J-10Cs), J-16s and J-20s that they are getting wrong and undercounting, which is not unexpected because it's difficult for them to follow the latest numbers.
But at the same time, those three types -- J-10C, J-16, J-20 -- are also easily the most capable aircraft that the PLAAF have in service, and so undercounting those types by whatever percentage that they are, could be fairly said to be "concerning" for them.
In particular, for J-20, the idea of there being 24 J-20s in service is of course very different to the more recent numbers that we got evidence for (specifically the number of aircraft in 9th brigade as indicated by 622xx serial)....
Given the importance of J-20 as a 5th generation fighter, undercounting of J-20s (if the true number is say, 40, or 50, or especially if it's more say up to 60 at present, or more?) then estimates of 24 J-20s certainly would be wrong and potentially a deadly underestimate.
At the end of the day, I expect military forces to be more prudent with their estimates than us -- the fact that they give specific numbers instead of not even giving a "range" of in service aircraft, is a bit silly IMO.