AA with TIPS for FI

[tulak]

I’m sharing this in case others are interested. There’s been a few threads discussing TIPS lately. I’m using TIPS for all FI in my AA, to deal with SORR and as insurance against inflation. I’ve been wanting to model an AA to see what the data says, in case I’m fooling myself in thinking this is a good idea.

I’m also curious if anyone else has seen an AA modeled this way elsewhere and if so, please let me know. And feel free to share your thoughts on how I’m modeling this portfolio. I’m sure I’m missing something…

Background: I’m primarily an equities investor. I’ve been 95%+ many times in the past and currently in the high 80%. I want to maintain as high of an equity allocation as possible. I’m looking to retire in the next 5-10 years and I’ve been thinking of how to deal with SORR. My approach is simple: I’ll use a TIPS ladder for 8 years and the remaining balance will be invested in equities.

Questions I’m trying to answer:

1. Is this a good AA? Why not use a traditional 80/20 or 60/40 AA?
2. Why 8 years for the TIPS ladder? Why not 5 or 10 years?

There are more questions, but these are the two that I’m focused on right now.

The problem I have is, how do I model this?

The available tools are all based on standard AA for total return with rebalancing at some cadence (or not). There are no modeling tools that I’m aware of that allow you to model using TIPS as FI and equities for good years. There is no rebalancing in this model. You use TIPS on down years and equities on up years. Since I don’t know of any tools that model this, I modeled it myself in a spreadsheet.

Assumptions:
1. Portfolio is $1,000,000
2. WR is 4% (40k/year)
3. If the previous year was down, withdraw from TIPS, if up, withdraw from equities
4. Equities are invested 100% in S&P500
5. The TIPS ladder is built prior to retirement

Ideally, I’d be able to model the start date on a bunch of different years, but that’s hard. I don’t have the data or tools. So I took a short-cut: I picked 1966 as the start year and modeled from that year forward. I figure if it works for a 1966 start, it should be good for other years.

Using these assumptions, here’s what I found:

With standard allocations (I used FiCalc.app):
1. 100/0, fails in 1992
2. 80/20, fails in 1992
3. 60/40, fails in 1991

With a TIPS ladder:
1. 5 year tips ladder, fails in 1995 with a balance of $176,936
2. 6 years tips ladder, fails in 1996 with a balance of $123,732
3. 7 years tips ladder, fails in 1995 with a balance of $25,369
4. 8 year tips ladder, fails in 1993 with a balance of $68,215
5. 9 years tips ladder, fails in 1991 with a 1988 balance of $103,944 - last 3 years were TIPS
6. 10 year tips ladder, fails in 1990 with a 1986 balance of $129,599 - last 4 years were TIPS

The balances are inflation adjusted and for the equity part of the portfolio. #1, the 5 years TIPS ladder, failed in 1995 because $188,302 was needed and only $176.936 was remaining. 40k in 1966 is quite a bit more in the 90s (> 4x more). Gotta love inflation.

Some observations using a TIPS ladder:
1. The sweet spot is a 6 year ladder. I was planning for 8 years, and thinking about 10, but maybe this is overkill?
2. If your TIPS ladder is too long, then you’ll run out of equities and live off TIPS until you run out of equity money. This is what happened with a 9 and 10 year ladder.

How could I improve this model:
1. I assume a 40k/year constant WR for both TIPS/equities. The reality is that withdrawals from a TIPS is a barebones amount where withdrawals from equities is the desired amount. So using 40k/year desired amount, maybe TIPS are only 30k/year and the TIPS could be stretched-out further?
2. I don’t take into consideration buying new TIPS when equities are high to keep the ladder going.
3. I don’t take into consideration interest from TIPS.
4. I could modify the criteria on when to withdraw TIPS vs equities.
5. Use more than 1966 as a start year. Monte Carlo simulations, etc. Using only one data pattern is not ideal for modeling.

Going back to the questions I asked above:

1. Is this a good AA? Why not use a traditional 80/20 or 60/40 AA?

It looks like a reasonable alternative to an 80/20 or 60/40 AA. Using TIPS, I added a few years before the portfolio was depleted. I also have the benefit of inflation protection, which maybe is the reason this allocation performs better?

Also of note: with a 6 year TIPS ladder, there’s 30 year success. I wonder if this would have a 100% success if all years were tested?

1. Why 8 years for the TIPS ladder? Why not 5 or 10 years?

It looks like 8 years is too long. I may go with 6 years, since that would allow a larger allocation to equities. I need to think on this and maybe model some of the other bad start years.

There is an aspect of this approach that deserves mention: the amount allocated to FI is fixed, regardless of portfolio size. There is no rebalancing. For the modeling above, I used a one million portfolio with a 40k WR, since that's standard. But using this approach, if you allocate 40k/year for 6 years for the FI allocation, there is never more than 240k invested in FI. This means you could easily have a 90%+ equity allocation while still mitigating SORR if you have a portfolio > million (or two). I consider this a feature, since I'm not risk adverse and want a high equity allocation. But it's definitely not for everyone.

Anyways, this has been an interesting exercise for me and hopefully others find this useful. And as I mentioned earlier, any comments, etc, is appreciated. I’m sure there are aspects that I haven’t thought of, but right now, I’m feeling reasonably good with investing in TIPS for FI.

 

[Exchme]

There is nothing wrong with holding TIPS as your fixed income, many people do that for at least some of their bonds. Note that when you have to sell TIPS prior to maturity, they can lose money due to unexpected inflation, just like nominal bonds. I don't have the numbers in front of me, but recall seeing numbers like nominal bonds lost 16% in 2022 and TIPS lost 14%. You are only guaranteed your original contract return if held to maturity, which your scheme won't let you do.

When you add all this complexity of how you plan to manage your portfolio, I'm not aware of any consumer grade modeling software that would let you do this. (How you do model 1966 when TIPS only started being offered in 1997?) I'm also not aware of any evidence it would increase your returns, other than by chance.

Note that there are many failure scenarios possible, continually testing alternatives vs. 1966 will optimize your strategy for conditions of 1966. If you look at the longer data series like Shiller's data, there were other bad sequences. For instance, sequences that started in the early 1900's were quite bad as there were doldrums from 1905-1921 plus the Great Depression. That needed a completely different strategy than 1966 and of course an endless variety of other patterns can and will exist in the future as well. For novel asset management strategies, you would have to examine a variety of approaches like Monte Carlo and hybrid approaches where randomly selected stretches of historical data are used.

So the question is why do you think this particular idea would help vs. a simpler approach like holding a fixed asset allocation with rebalancing or something like what Kitces advocates of a rising glidepath? Remember that lots of smart people have come before and looked at all kinds of schemes. If something similar to this was a robust improvement, you probably would have read about it.

 

[tulak]

There is nothing wrong with holding TIPS as your fixed income, many people do that for at least some of their bonds. Note that when you have to sell TIPS prior to maturity, they can lose money due to unexpected inflation, just like nominal bonds. I don't have the numbers in front of me, but recall seeing numbers like nominal bonds lost 16% in 2022 and TIPS lost 14%. You are only guaranteed your original contract return if held to maturity, which your scheme won't let you do.

Agree, which is why TIPS should be held to maturity.

When you add all this complexity of how you plan to manage your portfolio, I'm not aware of any consumer grade modeling software that would let you do this. (How you do model 1966 when TIPS only started being offered in 1997?) I'm also not aware of any evidence it would increase your returns, other than by chance.

You don't need TIPS to go back prior to 1997 to see how they would work in a model, assuming you always hold TIPS to maturity and TIPS nominal yield = CPI (only the inflation adjustment). CPI goes back to 1913, so there's plenty of historical data to get a feel for how TIPS would perform if only accounting for inflation.

Note that there are many failure scenarios possible, continually testing alternatives vs. 1966 will optimize your strategy for conditions of 1966. If you look at the longer data series like Shiller's data, there were other bad sequences. For instance, sequences that started in the early 1900's were quite bad as there were doldrums from 1905-1921 plus the Great Depression. That needed a completely different strategy than 1966 and of course an endless variety of other patterns can and will exist in the future as well. For novel asset management strategies, you would have to examine a variety of approaches like Monte Carlo and hybrid approaches where randomly selected stretches of historical data are used.

Yes, that's the hard part. I picked a known bad year to see how it would work. I should model other bad years, but as I mentioned, that's not easy. Modeling 1966 was a couple hours of work. Building a model to test all years from 1871 (well, 1913 since I need CPI data) is a bigger project.

So the question is why do you think this particular idea would help vs. a simpler approach like holding a fixed asset allocation with rebalancing or something like what Kitces advocates of a rising glidepath? Remember that lots of smart people have come before and looked at all kinds of schemes. If something similar to this was a robust improvement, you probably would have read about it.

This model is basically Kitces' use of a bond tent, using TIPS for bonds. Using TIPS eliminates inflation risk and guarantees minimum income. It also has hints of Kitches' rising glidepath approach, since this approach allows for a high equity allocation that increases with time, since the TIPS allocation is a fixed amount that never grows and no rebalancing.

I'm not trying to reinvent the wheel, only see if there's a better way using TIPS instead of traditional bonds. And to answer your question directly: based on one run starting in 1966 it does do better than a fixed AA with rebalancing. It might be cherry-picking, but maybe not (tbd). But this approach is easier and better suited for how I invest.

As for why it's not recommended by others, I would say that it depends who you ask. Bernstein advocates the use of TIPS for his liability matching portfolios (LMP). I think LMP is overly conservative, which is why I don't use that approach, except for a fixed number of years to get beyond SORR (which is really Kitces' bond tenting with TIPS). And so I don't misrepresent Bernstein, his ideal vehicle for LMP is an inflation indexed SPIA, but since those don't exist - beyond SS - the alternative is to use a ladder of TIPS.

I appreciate the response. When I posted this here, I didn't expect to get much interest, since most are beyond the point of thinking about SORR. It would be a better post for bogleheads, but I don't go there often, so figured I'd post here and see what happens.

 

[GravitySucks]

I appreciate the response. When I posted this here, I didn't expect to get much interest, since most are beyond the point of thinking about SORR. It would be a better post for bogleheads, but I don't go there often, so figured I'd post here and see what happens.

I think anyone under 70 or 75 and in decent health should worry about SORR. Just because one had been retired 10 years doesn’t mean your out of the woods. Still lots of time left to worry about.

Glad your model worked out as I’m doing close to the same.