Having calculated the 1Y stress scenario at certain dates on the Euro stoxx 50 series, I realise that it jumps during June 2017. As at 31/5/2017 I get 0.347660613, and as at 30/06/2017 I get 0.663384825.

Am I right in saying that the KID would show at the end of May that there is a stress risk of losing 65.23% over one year, while the KID at the end of June would show a stress risk of losing 33,66% over one year ?

Firstly, can someone indicate if my calculation is terribly wrong ?

Secondly, would financial advisors, clients, regulators be aware of the limitations of showing these scenario statistics at one single date, and without some interpretation ?

Regards.

  • If you know it, expand on why the value jumps in your code please. It is unhelpful to effort spent to hunt down the cause your own workings. If you provide the judged reason for the jump in value, I will run a cross check on this tomorrow and post my own. – James Spencer-Lavan Oct 10 '17 at 18:57
  • On 24 June 2016 the index lost 9.01%, which increased the 21 day volatility for the periods including that date (24/6/16 to 25/7/16) to above 0.025 . The observation points that look up these dates over 1Y show this increased value of volatility. From 26/6/17, one year after these observations, no other high volatility points are observed over 1Y, and 1Y stressed volatility returns to under 0.014, more often 0.011 to 0.01, even below 0.01 at end September. Since the 1Y scenario calculation is based on the 1Y stressed volatility, it is only logical that it improves as the volatility decreases. – Christian63 Oct 11 '17 at 7:58
  • to be honest it is not clear to me the procedure that you follow. For the determination of Wσ_{s}, as r_{t} do you use the historical returns of RHP (1year in your case) or do you use the historical data of (at least) last2-year period (minimum daily data requirement – Peter92 Oct 12 '17 at 15:12
  • @Peter92 to answer your question: For the 1 year stress scenario, for each day of the most recent year I calculate the volatility, using the excel formula for STDEV Pearson looking up the last 21 daily performance results (eg D24=STDEV.P(C3:C24) where C3=LN(B3/B2), C4=LN(B4/B3),…C24=LN(B24/B23)). The 1Y stress volatility is the value of volatility corresponding to the 99th percentile of all the 21 day volatility calculations over that last year. Hope this helps. (edited from previous answer) – Christian63 Oct 16 '17 at 7:06

I am a bit late to this, but I believe that your calculation is incorrect based on your response to the questions in the comments. The stress volatility used for the stress scenario should be calculated from rolling 21-day volatilities for your entire 5-year data sample (see PRIIPs stress scenario 2 calculation period of calculation ) and not just the most recent year. So, for example, in my calculations of the stress volatility used for the 1 year calculation, the stress volatility does not change between May and June 2017. Using only the most recent year would make the estimate of the stress scenario much more volatile.

In spite of that, I think you are right to have concerns regarding the interpretation of these figures. My own calculations shows that the stress scenario outcomes remain subject to wild changes from one month to the next. I constructed a chart from my calculations for the Euro Stoxx 50. The chart clearly shows large step changes as periods of major stock market volatility move in and out of the 5 year data sample window. For example, there is a large change in 2014 as the 2008-2009 volatility moves out of the sample. Category 2 Stress Calculation Chart - Euro Stoxx 50

  • Thanks for your comment. The regulation is not very clear on the lookback period but after some discussions I had already corrected my perspective. I now use 5 year data to follow stress volatility. My result looks similar to yours, although you dig further back in history! – Christian63 Mar 15 at 18:01
  • No problem. Yes, the regulations are not very clear - that's why I wanted to make sure that this question had a more definitive answer in case people stumbled across it. I also wanted to see if I got similar results to others, so thanks for confirming that! – Tim Wilding Mar 15 at 18:22
  • I do confirm that my graph looks exactly like yours, but only back to Feb 2003. I am also able to superimpose the other 3 scenarios. How do you find data for the index prior to its launch on Feb 1998 ? – Christian63 Mar 21 at 10:33
  • I used data from FactSet and didn't even realise it wasn't launched until Feb 1998. It's clear that history has been calculated for the index right back to 1987 - see boerse.de/historische-kurse/Euro-Stoxx-50/… – Tim Wilding Mar 21 at 15:31

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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