How to Make Switching Work
Posted: Sun Aug 17, 2003 10:20 am
I have tried continually to make switching work and I have been continually frustrated: until now. By switching, I am referring to shifting allocations based upon some measure of value such as P/E10. Repeated attempts, most often reported by BenSolar, have shown very little improvement in the Historical Database Rate. I was making another attempt, this time based upon TIPS (Treasury Inflation Protected Securities), when I figured out what we have been doing wrong. I know why we have failed in the past.
I was using a Safe Withdrawal Rate switching threshold. I had just failed to make 2.8% (yield to maturity) long-term TIPS improve the Historical Database Rates via switching when it occurred to me that the worst case dividend yield in the relevant period was 3%. That's right. My alternative investment did not produce as much interest as stocks provided from dividends alone.
In the past, whenever we have attempted dynamic allocation (i.e., shifting allocations), we have replaced stocks with investments that return less than the dividend yield of the stocks. For stocks to succeed and for our attempts to fail, it has been sufficient for stocks to regain their initial prices within the life spans of the portfolios that we are examining. Usually, that means 30 years or more. Failure has been guaranteed. To the extent that stocks yield more than our substitute investments, we will still fail even when stocks decline somewhat.
Making Use of the Historical Database
The Historical Database relates to stock investments that are no longer available. Until recently, stocks in the S&P 500 have yielded at least 3% and, often, much more. Our tools for examining the Historical Database are limited in how they treat investments other than stocks. The alternatives are treated strictly as short-term trading vehicles with a turnover no longer than one year. As such, we cannot use our tools directly to handle more reasonable strategies such as holding on to a long-term bond (at a favorable rate) until stocks become attractive. As far as our tools are concerned, any long-term bond vanishes forever after a single year.
We are limited in what we can do to introduce realism into Historical Sequence calculators. Today's Safe Withdrawal Rates are certainly less than in the past because dividend yields are down substantially. Dividend yields have provided a floor that has supported the safety of withdrawals. That support is no longer available. A direct comparison with Historical Database Rates is highly misleading.
What I am now doing is looking at historical parallels with today with differences clearly identified. In all cases I am looking for cause and effect relationships to support conclusions. More so now than in the past, the kinds of errors associated with any prediction are clearly visible.
Here is my Historical Parallel
I am interested in using today's TIPS to ride out the market until stocks become attractive once again. Today's TIPS yield 2.8% or so to maturity (maximum) and they last just under 30 years. Today's stocks provide a very low interest rate, roughly 1% less than today's TIPS.
My parallel is to use 4.1% 30-year TIPS when looking at the historical database. That is close to 1% more than stocks returned during the historical period.
Both 4.1% TIPS and stocks consistently yielding 3% and more have occurred in the past. Neither is outside the range of possibilities. Neither exists today.
Notionally, if 4.1% TIPS plus switching could have produced a 5% safe Historical Database Rate that lasted 50 years or more, then today's 2.8% TIPS can reasonably be expected to produce a 4% Safe Withdrawal Rate at similar levels of safety over a similar time period. (Although the Historical Database Rate is reported to be something very close to 100%, the actual level of safety is limited by a variety of factors, including many that apply most of the time.)
Notice that I have phrased the problem in terms that are both meaningful and solvable. For example, the 4.1% 30-year TIPS of the past would have had a 5% Safe Withdrawal Rate over 43 years assuming that they could be replenished. That is highly unlikely. But there would still have been up to 30 years before having to make an allocation shift. Similarly, it may or may not be possible to replace today's 2.8% TIPS in the future. However, there will be a long time before it is necessary to replace them.
Test Conditions
I identified all of the years with Safe Withdrawal Rates less than 6.0% from my previous post From Intrinsic Value. I did not add the 1% ( 0.74% to 1.36%) correction. Those years are 1928, 1929, 1930, 1936, 1937, 1939, 1946, 1955, 1956, 1957, 1959-1974. In addition, please note that 1931, 1938, 1940 and 1941 have asterisks, which indicate a sharp cut in the dividend yield within two or three years. When using the tables, use the revised headings: year, P/E10, dividend yield, safe withdrawal rate, historical database rate (80% stocks).
http://nofeeboards.com/boards/viewtopic.php?t=1220
The longest wait until the Safe Withdrawal Rate climbed above 6.0% was 16 years. That would have been in 1959.
TIPS at a 4.1% interest rate will provide a steady stream of 8.645% of an initial balance for 16 years before falling to a zero balance. This includes a return of capital (i.e., the drawing down principal) and it does not include any capital gains or losses. All amounts are adjusted for inflation.
To produce 5% of a portfolio's total initial balance for 16 years, a 4.1% TIPS account balance would fall to 80.20% of its initial value. To maintain the same withdrawal amount, the withdrawal rate at 16 years is 6.234% of the account balance at 16 years. (That is, 5% / 0.8020 = 6.234%.)
After the initial 16-year period, I switched to 80% stocks and 20% commercial paper. I used the CPI for inflation adjustments. I assumed 0.20% expenses. I assumed an initial balance of $1000 and made annual withdrawals of $62.34. I then determined the number of years that I could continue to make withdrawals according to the Historical Database. I used the FIRECalc for my calculations. http://capn-bill.com/fire/
I then determined the portfolio life span for all of the 1921-1980 portfolios that had Safe Withdrawal Rates of 6.0% or more. This tells us how many years that we could extend our retirement portfolio by waiting 16 years until we got a good buying opportunity. Anything bigger than 27 years would be an improvement over using TIPS alone. Assuming that the 4.1% TIPS could be replaced, TIPS by themselves would provide an identical income stream for exactly 43 years. (Remember that the income stream is 5% of the portfolio's initial balance.)
Success
Excluding the years with asterisks as well as those with Safe Withdrawal Rates below 6.0%, failures occurred in the following years: 1934 failed after 34 years, 1958 failed after 38 years, 1935 failed after 60 years and 1927 failed after 65 years. There were no other failures within the 80 years possible. (I.e., 1921 + 80 = 2001, which is the last entry in the FIRECalc.)
For the years with asterisks, only 1931 should be excluded automatically. It had a 5.6% Historical Database Rate. The Historical Database Rate for 1940 was 6.0%. The Historical Database Rate for 1938 was 6.6%. The Historical Database Rate for 1941 was 7.0%.
These were the failures: 1940 failed after 30 years, 1938 failed after 37 years and 1941 failed after 43 years.
We were able to extend the portfolio's life span more than 27 years every time by waiting for a good buying opportunity. In addition, we did not have to worry about running out of replacements for the TIPS that we did own.
Our earliest failure was after 46 years (from waiting 16 years and then using the additional 30 years for 1940). That data point had an asterisk, which might have influenced real-life decision-making. The second failure, with no asterisk, was after 50 years (from waiting 16 years and then using the additional 34 years for 1934).
Inferences
As indicated earlier, neither 4.1% TIPS nor an S&P 500 index yielding 3% (or more from dividends) is available today. However, 2.8% long-term TIPS are available on the secondary market. They should be able to provide a sufficient income stream while waiting for the Safe Withdrawal Rates of stocks to rise above 5%. That income stream would be close to 4% of one's initial portfolio balance. The portfolio should last at least 46 to 50 years.
We can extend our portfolio life span by using TIPS and our newly developed Safe Withdrawal Rate calculations and we can withdraw 4% of our initial portfolio balance in today's market.
Have fun.
John R.
I was using a Safe Withdrawal Rate switching threshold. I had just failed to make 2.8% (yield to maturity) long-term TIPS improve the Historical Database Rates via switching when it occurred to me that the worst case dividend yield in the relevant period was 3%. That's right. My alternative investment did not produce as much interest as stocks provided from dividends alone.
In the past, whenever we have attempted dynamic allocation (i.e., shifting allocations), we have replaced stocks with investments that return less than the dividend yield of the stocks. For stocks to succeed and for our attempts to fail, it has been sufficient for stocks to regain their initial prices within the life spans of the portfolios that we are examining. Usually, that means 30 years or more. Failure has been guaranteed. To the extent that stocks yield more than our substitute investments, we will still fail even when stocks decline somewhat.
Making Use of the Historical Database
The Historical Database relates to stock investments that are no longer available. Until recently, stocks in the S&P 500 have yielded at least 3% and, often, much more. Our tools for examining the Historical Database are limited in how they treat investments other than stocks. The alternatives are treated strictly as short-term trading vehicles with a turnover no longer than one year. As such, we cannot use our tools directly to handle more reasonable strategies such as holding on to a long-term bond (at a favorable rate) until stocks become attractive. As far as our tools are concerned, any long-term bond vanishes forever after a single year.
We are limited in what we can do to introduce realism into Historical Sequence calculators. Today's Safe Withdrawal Rates are certainly less than in the past because dividend yields are down substantially. Dividend yields have provided a floor that has supported the safety of withdrawals. That support is no longer available. A direct comparison with Historical Database Rates is highly misleading.
What I am now doing is looking at historical parallels with today with differences clearly identified. In all cases I am looking for cause and effect relationships to support conclusions. More so now than in the past, the kinds of errors associated with any prediction are clearly visible.
Here is my Historical Parallel
I am interested in using today's TIPS to ride out the market until stocks become attractive once again. Today's TIPS yield 2.8% or so to maturity (maximum) and they last just under 30 years. Today's stocks provide a very low interest rate, roughly 1% less than today's TIPS.
My parallel is to use 4.1% 30-year TIPS when looking at the historical database. That is close to 1% more than stocks returned during the historical period.
Both 4.1% TIPS and stocks consistently yielding 3% and more have occurred in the past. Neither is outside the range of possibilities. Neither exists today.
Notionally, if 4.1% TIPS plus switching could have produced a 5% safe Historical Database Rate that lasted 50 years or more, then today's 2.8% TIPS can reasonably be expected to produce a 4% Safe Withdrawal Rate at similar levels of safety over a similar time period. (Although the Historical Database Rate is reported to be something very close to 100%, the actual level of safety is limited by a variety of factors, including many that apply most of the time.)
Notice that I have phrased the problem in terms that are both meaningful and solvable. For example, the 4.1% 30-year TIPS of the past would have had a 5% Safe Withdrawal Rate over 43 years assuming that they could be replenished. That is highly unlikely. But there would still have been up to 30 years before having to make an allocation shift. Similarly, it may or may not be possible to replace today's 2.8% TIPS in the future. However, there will be a long time before it is necessary to replace them.
Test Conditions
I identified all of the years with Safe Withdrawal Rates less than 6.0% from my previous post From Intrinsic Value. I did not add the 1% ( 0.74% to 1.36%) correction. Those years are 1928, 1929, 1930, 1936, 1937, 1939, 1946, 1955, 1956, 1957, 1959-1974. In addition, please note that 1931, 1938, 1940 and 1941 have asterisks, which indicate a sharp cut in the dividend yield within two or three years. When using the tables, use the revised headings: year, P/E10, dividend yield, safe withdrawal rate, historical database rate (80% stocks).
http://nofeeboards.com/boards/viewtopic.php?t=1220
The longest wait until the Safe Withdrawal Rate climbed above 6.0% was 16 years. That would have been in 1959.
TIPS at a 4.1% interest rate will provide a steady stream of 8.645% of an initial balance for 16 years before falling to a zero balance. This includes a return of capital (i.e., the drawing down principal) and it does not include any capital gains or losses. All amounts are adjusted for inflation.
To produce 5% of a portfolio's total initial balance for 16 years, a 4.1% TIPS account balance would fall to 80.20% of its initial value. To maintain the same withdrawal amount, the withdrawal rate at 16 years is 6.234% of the account balance at 16 years. (That is, 5% / 0.8020 = 6.234%.)
After the initial 16-year period, I switched to 80% stocks and 20% commercial paper. I used the CPI for inflation adjustments. I assumed 0.20% expenses. I assumed an initial balance of $1000 and made annual withdrawals of $62.34. I then determined the number of years that I could continue to make withdrawals according to the Historical Database. I used the FIRECalc for my calculations. http://capn-bill.com/fire/
I then determined the portfolio life span for all of the 1921-1980 portfolios that had Safe Withdrawal Rates of 6.0% or more. This tells us how many years that we could extend our retirement portfolio by waiting 16 years until we got a good buying opportunity. Anything bigger than 27 years would be an improvement over using TIPS alone. Assuming that the 4.1% TIPS could be replaced, TIPS by themselves would provide an identical income stream for exactly 43 years. (Remember that the income stream is 5% of the portfolio's initial balance.)
Success
Excluding the years with asterisks as well as those with Safe Withdrawal Rates below 6.0%, failures occurred in the following years: 1934 failed after 34 years, 1958 failed after 38 years, 1935 failed after 60 years and 1927 failed after 65 years. There were no other failures within the 80 years possible. (I.e., 1921 + 80 = 2001, which is the last entry in the FIRECalc.)
For the years with asterisks, only 1931 should be excluded automatically. It had a 5.6% Historical Database Rate. The Historical Database Rate for 1940 was 6.0%. The Historical Database Rate for 1938 was 6.6%. The Historical Database Rate for 1941 was 7.0%.
These were the failures: 1940 failed after 30 years, 1938 failed after 37 years and 1941 failed after 43 years.
We were able to extend the portfolio's life span more than 27 years every time by waiting for a good buying opportunity. In addition, we did not have to worry about running out of replacements for the TIPS that we did own.
Our earliest failure was after 46 years (from waiting 16 years and then using the additional 30 years for 1940). That data point had an asterisk, which might have influenced real-life decision-making. The second failure, with no asterisk, was after 50 years (from waiting 16 years and then using the additional 34 years for 1934).
Inferences
As indicated earlier, neither 4.1% TIPS nor an S&P 500 index yielding 3% (or more from dividends) is available today. However, 2.8% long-term TIPS are available on the secondary market. They should be able to provide a sufficient income stream while waiting for the Safe Withdrawal Rates of stocks to rise above 5%. That income stream would be close to 4% of one's initial portfolio balance. The portfolio should last at least 46 to 50 years.
We can extend our portfolio life span by using TIPS and our newly developed Safe Withdrawal Rate calculations and we can withdraw 4% of our initial portfolio balance in today's market.
Have fun.
John R.