I am analyzing past financial crises, comparing the correlation structure amongst 20 broad market indices (equities, FX, credit, commodities) during different crises using amongst other things PCA and random matrix theory.

For every crisis I want to compare the pre-crises matrix to the crisis matrix. It is normally quite easy to identify the start date of a crisis given the discontinuous jumps but it is much harder to nail down the end date of a crisis. Everything I think of seems very subjective.

Anyone have any ideas of a more objective way to identify the end dates of past crises?


  • $\begingroup$ very very hard to define an 'end'. Possible candidates that spring to mind include short term unsecured bank funding levels / central bank policy regimes (loosening vs tightening), principally as measures of presence / absence of liquidity being a governing feature of crises. $\endgroup$ – Mehness Jul 3 '18 at 14:51
  • $\begingroup$ perhaps the evolution of your derived components themselves over rolling windows are the best indicator of regime change :) $\endgroup$ – Mehness Jul 3 '18 at 15:01
  • $\begingroup$ I think you should look at stock prices (especially prices of financial stocks). XLF for example reached a minimum in late feb or early may 2009 and then recovered strongly. The crisis was over. $\endgroup$ – noob2 Jul 3 '18 at 19:36

You could always look at published recession indicators, such as the one below.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.