Comment

This most recent in the series of Claessens, Kose, and Terrones papers continues tomine the extensive and useful data set they have constructed bringing together financial and real economic variable for a broad range of countries. This effort examines three aggregate financial time series—real domestic credit, real house prices, and real equity prices— for 21 countries, quarterly, over the past 50 years. That is, they start their journey with something like 12,000 separate observations in their vault. With this vast store of information, it is unsurprising that this is a paper about data, not the economic implications. The authors spend their time here characterizing the frequency, duration, and amplitude of the downturns and recoveries that are present in the time series. That said, the minute you start reading the paper, you encounter an eye-popping number: 470. The authors identify more than 470 financial cycles: 473 downturns and 480 upturns, that is, roughly 22 cycles per country in their data set. The natural question is, What are we to make of this apparently enormous number of financial cycles? With that as motivation, I will organize my comments into two parts: what the authors do and what is missing. Starting with what they do, I have already mentioned the data and the results. Going into a bit more detail, then, I will briefly discuss data concepts before turning to the methods used by Claessens et al. to identify the cycles.