A TIPS Ladder Survey

I built a series of calculators to learn about TIPS Ladders. They are modified versions of my Deluxe Calculators V1.1A08 (or V1dot1A08) that I refer to as Deluxe Calc V1dot1A08P01 through Deluxe Calc V1dot1A08P06.

[I use the letter P in honor of Peteyperson, a prominent participant of several discussion boards, who has drawn our attention to TIPS Ladders and how they can help us in the practical aspects of putting a portfolio together.]

With each of these calculators, you specify the ladder length (i.e., number of years) and one threshold, which is a percentage of your initial bond allocation. The calculators use this threshold to take a portion out of your initial bond amount and to put it into your stock portfolio. [In one instance, I transfer money out of stocks and into bonds.] The percentage transferred is 1/(length of your TIPS ladder). I have included a guard to prevent any transfer that would deplete the bond account entirely. The number of transfers can exceed the number of years in the ladder because interest payments add to the size of the bond account.

I have restricted my changes to conditions without any portfolio rebalancing.

I have not included any provision to replenish the bond account.

These simplifications make it easy to interpret results. This leads us in the right direction, but it does not present us with us a final strategy.

Results with Deluxe Calculator V1dot1A08P01

The calculator version ending with P01 compares the total of dividends and interest with the threshold. [The threshold amount is adjusted each year to match inflation.] If the dividends plus interest falls below the threshold, money is transferred from bonds to stocks.

The intention of this transfer is to maintain your stock holdings when your account is facing rough times. This is a direct way to avoid selling stock shares when prices have declined. Any shortfall in income that is not covered by dividends plus interest comes out of your bond holdings. You transfer the entire principal amount that matures in a particular year from bonds to stocks.

This was the most complex condition examined. There was a slight advantage versus not making any transfers (including comparisons with maintaining a constant allocation by rebalancing). There seemed to be a best threshold at 5%. You seemed to be best off when the number of years (i.e., the ladder length) is very large, typically 30 years or longer.

This points us toward making many small transfers from bonds to stocks. In practical terms, when a ladder length of 20 years is better than 10 years, you would take one-half of the principal amount from maturing bonds of a particular year and put it back into the ladder. You would transfer the other half into stocks.

My decision to combine principal and interest had the side effect of distorting my TIPS data. I relied on the Ibond data instead.

This distortion is caused by calculator limitations. There is only one line for reporting interest. Any principal adjustment (to match inflation) is included with the interest. Since the principal amount of TIPS falls during times of deflation, the amount of interest reported is low when making the threshold comparison. [The TIPS amount at maturity is never less than the original amount in nominal dollars, but the calculator does not go into that much depth.]

With Ibonds the principal never falls below zero.

I mention the following as a caution to those who would consider modifying earlier versions of my calculator or, possibly, even the original Retire Early Safe Withdrawal Calculation, Version 1.61 dated November 7, 2002, which is the original foundation for all of my calculators. The original Ibond algorithm had an error. It placed a floor of zero dollars to the combination of a principal amount and the interest coupon. In fact, the floor of zero dollars applies only to the principal amount. When overall prices decline, Ibonds retain their full principal amount in terms of nominal dollars and they always pay the full amount of the interest coupon. I have corrected the Ibond algorithm in my most recent calculators.

Results with Deluxe Calculator V1dot1A08P02

I was confused by the behavior with my original algorithm. It did not seem to lead to easy interpretation.

I changed the algorithm from using dividends plus interest to P/E10. Part of my rationale had to do with previous studies that examined how Historical Surviving Withdrawal Rates were related to (initial) dividend yields. It turned out that the percentage earnings yield 100/[P/E10] did better. Historically, there have been several years with dividend cuts. The relationship with earnings yield turns out to be more reliable. Earnings yield automatically takes the quality of dividends into account.

I made a timing error when I constructed the Deluxe Calculator V1dot1A08P02. I correct this when I put together version P03.

Results with Deluxe Calculator V1dot1A08P03

I sold bonds and bought stocks whenever P/E10 was above a threshold. This is similar to having dividends fall below a threshold. This is similar to my original comparison.

What I found was that the best threshold was zero. That is, I should transfer money out of bonds and into stocks, regardless of valuations. Conditions favored very long ladders, typically 30 to 50 years. Conditions generally favored starting with a small stock allocation (around 20% and a large bond allocation).

Of significance, moving small amounts from bonds into stocks over an extended time period ended up being slightly better than having a fixed allocation with rebalancing. [A fixed allocation with rebalancing was better than a fixed allocation without rebalancing.]

My data do not allow me to draw the strong positive conclusion that you should move small amounts from bonds to stocks instead of using a fixed allocation with rebalancing. But they are sufficient for me to refute the negative conclusion that you are necessarily better off by sticking with a fixed allocation and rebalancing.

These findings are helpful for those who are reluctant to own stocks. During times that stress retirement portfolios, you are at least as well off by starting with a large bond (i.e., TIPS and/or Ibonds) allocation (around 80%) and gradually buying stocks (about 2% to 4% of your initial portfolio amount plus inflation annually) as bonds mature.

Because of volatility, it can take a lot of time before you can guarantee that a stock portfolio has grown. Starting with a small stock allocation and adding to it very gradually assures you that your stocks have grown before you allow them to dominate your portfolio.

Results with Deluxe Calculator V1dot1A08P04

I changed the comparison to whether P/E10 is below the threshold.

This resulted in a best threshold around 12 or 13. This is similar to the thresholds that are optimal when switching stock allocations according to P/E10.

This brings the lowest Historical Surviving Withdrawal Rate above 4.6% (but below 4.7%), which is about half of the way between using fixed allocations (around 4%) and using switching (above 5%).

Considering that this algorithm never replenishes any of the bond holdings, it does a great job. It is similar to switching.

Results with Deluxe Calculator V1dot1A08P05

This is the condition in which I sold stocks to buy bonds.

I examined buying more bonds by selling stock when P/E10 was above threshold. This shares the same general idea as buying stocks from the bond portfolio when P/E10 is below a threshold. But it is not the same.

I found that the best threshold is 22 or higher. The improvement was not as dramatic as before.

Results with Deluxe Calculator V1dot1A08P06

I went back to using dividends plus interest. This time, I added stock whenever dividends plus interest were above the threshold.

I had to use Ibonds for comparisons. A withdrawal rate of 4.7% produced no failures (but 4.8% did). I did not make direct comparisons with TIPS beyond noting that there was a substantial distortion during times of deflation.

In contrast to the initial comparison (when dividends plus interest were below threshold), this approach did well and it produced understandable results. This is similar to adding stock whenever P/E10 falls below a threshold.

The best threshold was around 9%, which is clearly in bargain territory.

The best ladder lengths were very short, around 2 to 3 years.

This tells us a simple story. Load up on stocks whenever they are dirt cheap (as indicated by huge dividends). Buy heavily.

[The short bond ladder makes it clear that you should purchase large quantities fast. Remember that the amount purchased is one over the length of the ladder. In this case, you would be selling one-third to one-half of your bond holdings to buy stocks.]

Interpretation

We need to change our statement about what kills retirement portfolios. It is selling stocks at depressed prices during the short-run that kills retirement portfolios.

Over longer periods of time, you are better off adjusting your allocations in accordance with the intrinsic value of your holdings. In our investigations into switching stock allocations, we found that P/E10 is the best among several good indicators of intrinsic value over the intermediate-term.

In this case, you should add to your stock holdings when P/E10 is low and/or dividend yields are high. You should reduce your stock holdings when P/E10 is high and/or dividend yields are low.

But you should look for good prices when you make your shorter-term changes. Allow yourself two or three years to get a reasonable stock price. Stay away from panic selling. Wait for prices to recover to a reasonable level, but possibly substantially lower than when you started.

What you are trying to do in the short-term is to get something close to an average stock price (when measured over one or two years). You are not demanding a good price. You are only avoiding a bad price. By waiting a year or two, if necessary, you are taking advantage of volatility in the market. You are waiting for prices to recover somewhat. Your are not demanding that they recover completely.

Have fun.

John Walter Russell
I wrote this on May 4, 2005.