“…For simplicity, in this section we only consider positive and negative policy shifts (i.e., δ ∈ {−1, 1}), since it is hard to identify the timing of "no policy change" (i.e., δ = 0). Because there was always a possibility of having a non-regular policy meeting especially in the 80's, looking only at the regularly held meetings does not fully capture the decision of the central banks to do nothing (Thornton, 2006;Kobayashi, 2009). It should be noted that the number of reversal policy changes would be lower if we included the events with δ = 0.…”