“…Our preference for K = 6 mainly reflects our prior (based on Schonhardt‐Bailey, 2013) that setting K too low would not allow us to identify a separate financial stability topic. We are hesitant to set K much higher, as many document sets are found to be best characterized by relatively few and broadly‐defined topics (for discussion and examples, see Blei et al, 2003; Maier et al, 2018; Sbalchiero & Eder, 2020; Istrefi et al, 2020).…”