Announcement

Collapse
No announcement yet.

Actuary on FIRE

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Actuary on FIRE

    For those of you who haven't found Actuary on FIRE through POF's Sunday best, I wanted to let you know it is definitely worth adding to your financial reading list.  My personal favorites are the posts on longevity and sequence of returns risk. How Sequence of Returns is discussed is different than how I have seen it presented before.  It really hits closer to how I think about it.  Hope you enjoy!

    https://www.actuaryonfire.com

     

  • #2
    Thanks for the heads up Dr. Mom.   I Read both articles.  The Sequence of returns risk is sobering. Wow.

    Comment


    • #3
      Thanks for sharing, Dr. Mom. Acuary on FIRE does a great job of explaining sequence of return risk. More importantly, he stressed the need to have a game plan for how to deal with it.

      Comment


      • #4
        The SOR post was intriguing. He also wrote a guest post on my site about longevity risk. I like how he framed the cost of living one more year in terms of how much more you need to save today.

        Comment


        • #5




          Thanks for the heads up Dr. Mom.   I Read both articles.  The Sequence of returns risk is sobering. Wow.
          Click to expand...


          it's weird how we must have subjectively known about SORR but minimized it, while comforting ourselves by focusing on trinity study, which told us what we wanted to hear.

           

          Comment


          • #6
            I thought my ears were burning! Many thanks for the kind comments, and stopping by the blog. I really appreciate that.  

             


            it’s weird how we must have subjectively known about SORR but minimized it, while comforting ourselves by focusing on trinity study
            Click to expand...


            What a good comment! The Trinity study considered data from 1926-2009 and 30 year retirement periods. All the argument has been around whether 30 years is long enough, and/or whether this is a sufficiently long historical period (no on both counts). But most seem to be over-looking the fact that by looking at retirements starting 1926-1980 you are only capturing 55 different sequences of returns. That's too small a data set.

            So I'm going to be extending my work to cover more historical periods *and* then scrambling up the order of those periods, so we can look at 100,000's of different sequences of returns. Stay tuned. Although this is starting to feel like more work than my PhD...

            Comment


            • #7


              So I’m going to be extending my work to cover more historical periods *and* then scrambling up the order of those periods, so we can look at 100,000’s of different sequences of returns.
              Click to expand...


              Do you believe scrambling historical periods provides some advantage over Monte Carlo simulations?
              Erstwhile Dance Theatre of Dayton performer cum bellhop. Carried (many) bags for a lovely and gracious 59 yo Cyd Charisse. (RIP) Hosted epic company parties after Friday night rehearsals.

              Comment


              • #8
                Builidng on CM's point, I think there are two worthwhile cases: (i) looking at historical 30 year periods and (ii) looking at Monte Carlo performance based on historical results. Scrambling historical periods doesn't add anything of value.

                Comment


                • #9


                  Do you believe scrambling historical periods provides some advantage over Monte Carlo simulations?
                  Click to expand...


                  They are two different things designed to answer two different questions. By taking different permutations of returns I'm just doing a statistical analysis of how sequence of returns impacts solvency. I agree that it provides no economic value or insight at all. It's purely a statistical way of isolating sequence of returns and answering the question:
                  If I keep the total return over the period constant, how could my solvency be impacted by the sequence of those returns?

                   

                  If I want economic insights into retirement outcomes then Donnie correctly points out you have two tools:

                  (1) historical periods

                  (2) Monte Carlo simulations of future economic regimes

                   




                  Builidng on CM’s point, I think there are two worthwhile cases: (i) looking at historical 30 year periods and (ii) looking at Monte Carlo performance based on historical results. Scrambling historical periods doesn’t add anything of value.
                  Click to expand...


                  Monte Carlo simulations don't achieve what I want, because they change everything at once, and therefore the total return over the period changes. I just want to isolate the sequence of return bit. Make sense?

                   

                   

                  Comment


                  • #10




                    Monte Carlo simulations don’t achieve what I want, because they change everything at once, and therefore the total return over the period changes. I just want to isolate the sequence of return bit. Make sense?
                    Click to expand...


                    Your point makes perfect sense. Monte Carlo takes away the zero-sum nature of scrambling the historical results.

                    I think your article takes a very difficult subject for the average investor to understand and makes it very understandable and shows people how it would relate to their personal retirement situation. Thank you for your insight and helping give me another way to explain this risk to my clients.

                    Comment


                    • #11
                      ERN dedicated 2 parts of his SWR series to sequence of returns risk for those interested in an academic look from a Ph.D. economist:

                       

                       

                      Comment


                      • #12




                        Monte Carlo simulations don’t achieve what I want, because they change everything at once, and therefore the total return over the period changes. I just want to isolate the sequence of return bit. Make sense?
                        Click to expand...


                        And this ladies and gentleman is why I like his blog so much.  I appreciate how he thinks about SORR very differently than anywhere else I have read.  I find great value in seeing how other people think about finance.  Thanks AOF!  Looking forward to your future posts!

                        Comment


                        • #13





                          Do you believe scrambling historical periods provides some advantage over Monte Carlo simulations? 
                          Click to expand…


                          They are two different things designed to answer two different questions. By taking different permutations of returns I’m just doing a statistical analysis of how sequence of returns impacts solvency. I agree that it provides no economic value or insight at all. It’s purely a statistical way of isolating sequence of returns and answering the question:
                          If I keep the total return over the period constant, how could my solvency be impacted by the sequence of those returns?

                           

                          If I want economic insights into retirement outcomes then Donnie correctly points out you have two tools:

                          (1) historical periods

                          (2) Monte Carlo simulations of future economic regimes

                           




                          Builidng on CM’s point, I think there are two worthwhile cases: (i) looking at historical 30 year periods and (ii) looking at Monte Carlo performance based on historical results. Scrambling historical periods doesn’t add anything of value.
                          Click to expand…


                          Monte Carlo simulations don’t achieve what I want, because they change everything at once, and therefore the total return over the period changes. I just want to isolate the sequence of return bit. Make sense?

                           

                           
                          Click to expand...


                          Do you feel keeping the total return the same makes up for the non independent nature of year to year returns? Which is also a problem with monte carlo results and why they so over and under shoot by large amounts what happens in reality.

                          Comment


                          • #14





                            Do you believe scrambling historical periods provides some advantage over Monte Carlo simulations? 
                            Click to expand…


                            They are two different things designed to answer two different questions. By taking different permutations of returns I’m just doing a statistical analysis of how sequence of returns impacts solvency. I agree that it provides no economic value or insight at all. It’s purely a statistical way of isolating sequence of returns and answering the question:
                            If I keep the total return over the period constant, how could my solvency be impacted by the sequence of those returns?

                             

                            If I want economic insights into retirement outcomes then Donnie correctly points out you have two tools:

                            (1) historical periods

                            (2) Monte Carlo simulations of future economic regimes

                             




                            Builidng on CM’s point, I think there are two worthwhile cases: (i) looking at historical 30 year periods and (ii) looking at Monte Carlo performance based on historical results. Scrambling historical periods doesn’t add anything of value.
                            Click to expand…


                            Monte Carlo simulations don’t achieve what I want, because they change everything at once, and therefore the total return over the period changes. I just want to isolate the sequence of return bit. Make sense?

                             

                             
                            Click to expand...


                            I don't think it makes sense because you are assuming that the historical results are the probability distribution as opposed to representative of the distribution.  In effect you are doing a Monte Carlo that pulls from an improper probability distribution.  I'm not sure what you mean about changing the total return over the period.  Using a Monte Carlo with the appropriate assumptions will get you to the place you want.

                            Comment


                            • #15
                              A question comes to mind when I think about the SORR.  Is it a true reflection of market performance to assume that the next year's return, or the next few years of returns has no relationship to what has come before?  The SORR analysis assumes that next year's return will be a random event.

                              In my mind, if the market has been down for the last couple of years, is next year's return truly a random event, or does the probability of an upward market return increase to higher than random after a few years of negative returns.  I would think it does, making a completely random SORR type of projection simply an interesting mathematical analysis, but one not based on the reality of future market returns being to some degree related to what has come in the preceding years.

                              Comment

                              Working...
                              X