5 ms·

How long is a 'long term investment'? (a brief analysis of S&P500 since 1950)

27 points by tomsaffell 18y ago

mixmax 18y ago
One of the flaws in this is that since he's looking back in time he doesn't include potential investments in companies that go bankrupt. This would shift the whole thing somewhat to the left. Interesting nonetheless.
- mattmaroon 18y ago
  I don't see how that matters. You can and maybe should just invest in SPDRs.
  - mixmax 18y ago
    OK, here is the explanation. Maybe it isn't as obvious as I thought. Let's say that you invest in 100 random stocks in 1978 and intend to keep them as a long term investment. In 2008 when you want to sell your stock it has on average risen by x%. But some of the companies you invested in have gone bankrupt, and thus these shares are worth nothing. This pulls your entire portfolio down by quite a bit. Now if you look back from 2008 instead of looking forward from 1978 you will see a different picture. If you pick 100 random stocks and see what their stockprice was in 1978 (which is what this guy seems to have done) you might expect to get the same result, but you don't. A lot of companies have gone bankrupt in those 30 years, and you don't include these in your back-looking portfolio. Yet, as we see these days, it is a real scenario that the company you have invested in will simply tank and your shares will be worth nothing. Over thirty years my guess is that 3 out of 100 companies will go bankrupt, meaning that bankruptcies alone diminishes your portfolios value by 3%. I think this is quite substabtial.
    - mattmaroon 18y ago
      I still don't understand. He's tracking the S&P 500 which changes over time. Have any companies gone bankrupt from that before being delisted? If so, they'd be factored in already. If not, then someone with SPDRs (had they existed) wouldn't have cared.
    - mattmaroon 18y ago
      Also bankruptcies count into your average return of x% if you're tracking the market as a whole. 3 companies will go bankrupt (-100%) but 1 that grossly outperforms the market (+1000% in a few years, for example) more than covers it. It's an average, that's what averages do.
    - tomsaffell 18y ago
      This is an interesting point. The analysis was done on the 'index value' of the S&P 500 (from Yahoo Finance), not on any individual equities. I'm trying to discover how long a 'long term investment' need be, if one invests in an S&P 500 index / ETF / iShare. (I'm no expert on the subtleties of those investment vehicles). My understanding is that they track (as best they can) the value of the index by investing in the stocks that compose the index. Therefore I guess that when company in the S&P 500 goes bust the effect that has on one's investment closely matches the effect it has on index value. I'll research that. Assuming this is true, I think we needn’t be concerned by the bankruptcy issue that you make. But please correct me if I'm wrong about that - I'm interested. I'll try to add TSR in the next analysis too.
      - mattmaroon 18y ago
        It's not interesting, it's just wrong. Even if the total market averages 10%, that's counting bankruptcies. It's not a median or a mode, it's a mean.
        mixmax 18y ago
        Matt, you're right. I thought he was looking at a portfolio of individual stock, not an index where bankruptcies are supposedly included. My bad...
      - mixmax 18y ago
        ahh, yes you're tracking S & P 500. Sorry, I probably read the aricle a little quickly... I don't know for sure, but I would think you're right in your assertion that the index incorporates the risk of companies failing.
mattmaroon 18y ago
He's missing dividends. The average S&P 500 dividend yield is 1.37% according to something I just found on Motley Fool. If that's accurate, that's pretty substantial. Too bad they didn't have SPDRs in 1950, he could have been much more accurate with the same amount of effort.
- tomsaffell 18y ago
  The analysis has now been updated to include dividends, and posted on my blog: http://saffell.wordpress.com/2008/10/26/does-timing-the-market-work/ http://saffell.wordpress.com/2008/10/26/does-timing-the-mark...
- eru 18y ago
  http://www.crossingwallstreet.com/archives/2006/03/sp_500_total_re.html http://www.crossingwallstreet.com/archives/2006/03/sp_500_to... has a chart of the total return. Or you can make your own at http://www.economagic.com/em-cgi/charter.exe/fedstl/trsp500 http://www.economagic.com/em-cgi/charter.exe/fedstl/trsp500 By the way more modern indices like the German DAX are already performance ones.
  - mattmaroon 18y ago
    Right, I just meant it would have been nice to see that incorporated into the variance.
- tomsaffell 18y ago
  Good point. I would like to analyze returns assuming re-invested dividends (TSR), which I will try to do in my next analysis. However, as I search around the web I'm having trouble finding enough data (SPDRs only go back to '93, and even the economagic data only starts in 1970, which isn't really enough data to meaningfully look at 'average' 30 year investments) Does anyone have access to data going further back? I will continue looking myself, maybe also looking at DJI data. In the spirit of hypothesis lead analysis, perhaps we (or I) should lay out hypotheses for how the TSR analysis will differ (Total Shareholder Return). Clearly the curves will be shifted to the right, probably by the ~1.37% that Matt mentions. But what about the variance? (which is more interesting IMO). I need to think about that..
marvin 18y ago
There are lots of interesting things to consider here. Thanks for putting this together, it set me thinking. The analysis doesn't consider investment over time - as far as I can tell, this graph shows the probability distribution of returns on a one-time investment after a period of years. The goal of investing long-term is to dampen the effect of investing at a (in retrospect) very good or very bad time. (In addition, of course, you need to put your money somewhere to get returns at all). Random oscillations can be dampened further by investing at regular intervals, disregarding what the stock price is at the moment: for instance, investing a smaller amount every month for five years (or even every month, period). It would be interesting to see how the probabilities of such an investment scheme fares against a savings account. You could even make two probability distributions: one for stock market returns and one for savings account returns, considering taxes where they apply. (In my part of the world, income from savings accounts is taxed annually while returns on stock is only taxed when the gains are realized, which affects returns over time substantially). Inflation affects all investments equally, and could be disregarded. Now that I think of it, maybe I should do this myself. This is exactly the kind of investment scheme I am betting on for the next 20 years, so it seems a bit irresponsible not to do this kind of thing beforehand.
- tomsaffell 18y ago
  Good thoughts. As you say, this is for a 'lump sum' investment, not investment over time, which has the benefit of dollar-cost-averaging. As I mentioned in my post, I may look at that next, probably in v3. The tax is is good point too, but I'll leave that off the table for the next version as people's tax situations vary, and it gets complex - maybe v5..
- gamble 18y ago
  Investing a lump sum by spreading it out over a period of time ('dollar cost averaging') has been shown to produce lower returns than making one large initial investment. Most experts regard it as a marketing gimmick used to ease nervous customers into investments.
  - comatose_kid 18y ago

27 points by tomsaffell 18y ago

One of the flaws in this is that since he's looking back in time he doesn't include potential investments in companies that go bankrupt. This would shift the whole thing somewhat to the left. Interesting nonetheless.

mattmaroon 18y ago
I don't see how that matters. You can and maybe should just invest in SPDRs.
- mixmax 18y ago
  OK, here is the explanation. Maybe it isn't as obvious as I thought. Let's say that you invest in 100 random stocks in 1978 and intend to keep them as a long term investment. In 2008 when you want to sell your stock it has on average risen by x%. But some of the companies you invested in have gone bankrupt, and thus these shares are worth nothing. This pulls your entire portfolio down by quite a bit. Now if you look back from 2008 instead of looking forward from 1978 you will see a different picture. If you pick 100 random stocks and see what their stockprice was in 1978 (which is what this guy seems to have done) you might expect to get the same result, but you don't. A lot of companies have gone bankrupt in those 30 years, and you don't include these in your back-looking portfolio. Yet, as we see these days, it is a real scenario that the company you have invested in will simply tank and your shares will be worth nothing. Over thirty years my guess is that 3 out of 100 companies will go bankrupt, meaning that bankruptcies alone diminishes your portfolios value by 3%. I think this is quite substabtial.
  - mattmaroon 18y ago
    I still don't understand. He's tracking the S&P 500 which changes over time. Have any companies gone bankrupt from that before being delisted? If so, they'd be factored in already. If not, then someone with SPDRs (had they existed) wouldn't have cared.
  - mattmaroon 18y ago
    Also bankruptcies count into your average return of x% if you're tracking the market as a whole. 3 companies will go bankrupt (-100%) but 1 that grossly outperforms the market (+1000% in a few years, for example) more than covers it. It's an average, that's what averages do.
  - tomsaffell 18y ago
    This is an interesting point. The analysis was done on the 'index value' of the S&P 500 (from Yahoo Finance), not on any individual equities. I'm trying to discover how long a 'long term investment' need be, if one invests in an S&P 500 index / ETF / iShare. (I'm no expert on the subtleties of those investment vehicles). My understanding is that they track (as best they can) the value of the index by investing in the stocks that compose the index. Therefore I guess that when company in the S&P 500 goes bust the effect that has on one's investment closely matches the effect it has on index value. I'll research that. Assuming this is true, I think we needn’t be concerned by the bankruptcy issue that you make. But please correct me if I'm wrong about that - I'm interested. I'll try to add TSR in the next analysis too.
    - mattmaroon 18y ago
      It's not interesting, it's just wrong. Even if the total market averages 10%, that's counting bankruptcies. It's not a median or a mode, it's a mean.
      - mixmax 18y ago
        Matt, you're right. I thought he was looking at a portfolio of individual stock, not an index where bankruptcies are supposedly included. My bad...
    - mixmax 18y ago
      ahh, yes you're tracking S & P 500. Sorry, I probably read the aricle a little quickly... I don't know for sure, but I would think you're right in your assertion that the index incorporates the risk of companies failing.

mattmaroon 18y ago

He's missing dividends. The average S&P 500 dividend yield is 1.37% according to something I just found on Motley Fool. If that's accurate, that's pretty substantial. Too bad they didn't have SPDRs in 1950, he could have been much more accurate with the same amount of effort.

tomsaffell 18y ago
The analysis has now been updated to include dividends, and posted on my blog: http://saffell.wordpress.com/2008/10/26/does-timing-the-market-work/ http://saffell.wordpress.com/2008/10/26/does-timing-the-mark...
eru 18y ago
http://www.crossingwallstreet.com/archives/2006/03/sp_500_total_re.html http://www.crossingwallstreet.com/archives/2006/03/sp_500_to... has a chart of the total return. Or you can make your own at http://www.economagic.com/em-cgi/charter.exe/fedstl/trsp500 http://www.economagic.com/em-cgi/charter.exe/fedstl/trsp500 By the way more modern indices like the German DAX are already performance ones.
- mattmaroon 18y ago
  Right, I just meant it would have been nice to see that incorporated into the variance.
tomsaffell 18y ago
Good point. I would like to analyze returns assuming re-invested dividends (TSR), which I will try to do in my next analysis. However, as I search around the web I'm having trouble finding enough data (SPDRs only go back to '93, and even the economagic data only starts in 1970, which isn't really enough data to meaningfully look at 'average' 30 year investments) Does anyone have access to data going further back? I will continue looking myself, maybe also looking at DJI data. In the spirit of hypothesis lead analysis, perhaps we (or I) should lay out hypotheses for how the TSR analysis will differ (Total Shareholder Return). Clearly the curves will be shifted to the right, probably by the ~1.37% that Matt mentions. But what about the variance? (which is more interesting IMO). I need to think about that..

marvin 18y ago

There are lots of interesting things to consider here. Thanks for putting this together, it set me thinking. The analysis doesn't consider investment over time - as far as I can tell, this graph shows the probability distribution of returns on a one-time investment after a period of years. The goal of investing long-term is to dampen the effect of investing at a (in retrospect) very good or very bad time. (In addition, of course, you need to put your money somewhere to get returns at all). Random oscillations can be dampened further by investing at regular intervals, disregarding what the stock price is at the moment: for instance, investing a smaller amount every month for five years (or even every month, period). It would be interesting to see how the probabilities of such an investment scheme fares against a savings account. You could even make two probability distributions: one for stock market returns and one for savings account returns, considering taxes where they apply. (In my part of the world, income from savings accounts is taxed annually while returns on stock is only taxed when the gains are realized, which affects returns over time substantially). Inflation affects all investments equally, and could be disregarded. Now that I think of it, maybe I should do this myself. This is exactly the kind of investment scheme I am betting on for the next 20 years, so it seems a bit irresponsible not to do this kind of thing beforehand.

tomsaffell 18y ago
Good thoughts. As you say, this is for a 'lump sum' investment, not investment over time, which has the benefit of dollar-cost-averaging. As I mentioned in my post, I may look at that next, probably in v3. The tax is is good point too, but I'll leave that off the table for the next version as people's tax situations vary, and it gets complex - maybe v5..
gamble 18y ago
Investing a lump sum by spreading it out over a period of time ('dollar cost averaging') has been shown to produce lower returns than making one large initial investment. Most experts regard it as a marketing gimmick used to ease nervous customers into investments.
- comatose_kid 18y ago