Author Topic: Pick your diversifiers wisely - rant against bonds (Read 4090 times)

vand · « **on:** October 12, 2024, 07:46:00 AM »

Bonds. F**king useless, aren't they?

Well... not quite, but yes, almost.

First, lets revisit why we may want to hold bonds - as a diversifying asset in a primarily stock-based portfolio, to dampen the volatility and to improve the risk-adjusted return characteristics of that portfolio - hence how the mighty 60/40 stock/bond portfolio - on edge of the supposed "efficient frontier" of return vs standard deviation - became the be all end all for how to construct a sleep-easy portfolio.

Bonds tend to permeate financial literature as the automatic and best balancing asset - how often have you heard someone asking "should I also hold some bonds" implicit in the question that the rest of the portfolio is made up of stocks? However, this assumption in general finance, is at best highly contestable, and for specific FI purposes, simply nothing more than dogmatic, incorrect, and dangerous.

Firstly, challenging the mighty 60/40 in general finance, it can be shown that other portfolio mixes can offer similar if not slightly better level of risk-adjusted return, for example a 75/25 stock/gold portfolio has beaten the 60/40 over the long term in absolute and risk adjusted terms. I won't go too much into detail on this as the datasets for different assets can be hard to find, and the starting year matters, so these numbers can be batted back and forth endlessly.

But what I really what to do here is drill down here in just how terrible bonds are as a diversifier for FI purposes, and how poor they are at improving safe withdrawal rates.

On PortfolioCharts I took a portfolio of a 100% US Large Cap Blend, which by itself offers a SWR of 3.8% over 30yr as a baseline.
I then added in the following asset in 5% increments to see what effect it had on the SWR:

Short Term US Treasuries (1-3Y)
Intermediate Term US Treasuries (3-10Y)
Long Term US Treasuries (10-30Y)
US REITs
Gold
Commodities

Here are the results:

I mean, the preponderance of bonds in common financial literature is pretty indefensible given these results, aren't they?

Whenever someone asks "should I hold some bonds in retirement" I would counter with "only if you already other diversifiers". Bonds, by themselves, are the WORST diversifying option you can make.

I, and others such as Tyler, have written about how well gold works as a diversifying asset and if you want to keep it simple then stock/gold portfolio is your best option.

But, I really hadn't realised just how poor bonds are relative to, well, just about everything else. By only considering stock/bond allocations you are restricting your SWR to this universe of historic outcomes...

I mean, if you knew better, why on earth would you do that? The best mix of stocks/bonds raises the SWR of the portfolio by 10.53% (3.8% to 4.2%), compared to the best mix of stocks/gold which raises the SWR of the portfolio by 39.47% (3.8 to 5.3%) - making gold almost 4 times as effective as bonds as a diversifying asset for this particular base scenario.

It's also very noticeable that the duration of bond doesn't really make any difference - in fact the short duration bonds produce a slightly higher SWR than intermediate and long duration, althought it may be a little as a rounding error... so much for the notion that you get paid more for higher duration.

What about PortfolioChart's own Global Withdrawal Rate Portfolio? That has a 20% allocation to bonds, doesn't it?
Yes it does, and if you are going to backtest more blends than just 2 assets then you will find a portfolio blended from more assets will be the optimal solution.. but you'll also need to crunch thousands of scenarios which I dont have the means to do. However I don't think it changes what I've said above, and as Tyler himself commented on his interview with Pensioncraft, if you had to pick just 1 diversifying asset, it should probably be gold... excellent interview btw, everyone should listen: https://pensioncraft.com/captivate-podcast/building-a-bulletproof-retirement-portfolio/

Lastly, I often hear on these pages "if your portfolio is large enough in relation to your living costs you might as well be 100% equities" - which I would greatly contest. Yes, it may give you higher upside potential if you have a sub-3% current withdrawal rate, but to what purposes? Just to be the richest person in the graveyard? Holding a more aggressive portfolio may mean it grows larger... but you'll never actually be able to enjoy that wealth. I'm not saying there there aren't reasons why you may want to die with sizeable wealth, but the reason usually given to remaining 100% equities do not hold up to logical cross-examination on a first principles basis.

Takeaways:

- In decumulation, Gold is the single best diversifier, not bonds. Gold should be your diversifier of choice, then add other assets in afterwards (including bonds) if you want
- Bonds are single worst asset class to use for diversifying a retirement portfolio. They should only be considered after you have added gold
- In accumulation, it doesn't matter as much, but even then blended stock/bond portfolio are only amongst the most efficient, not unarguably the most efficient
- REITs have also worked well, although the dataset for this may be more questionable - maybe @Tyler himeself can comment. Also REITs don't survive the backtests to make it into the Global Withdrawal Portfolio, so it may be squeezed out when considered from a wider range of scenarios

VanillaGorilla · « **Reply #1 on:** October 12, 2024, 10:40:49 AM »

I've done the same analysis many times and always come to the same conclusion: a 52 year timeframe is too small to be useful. The results are overfit to one or two cohorts: 1973 and maybe 2000.

I don't disagree with your conclusion that bonds are not very useful, historically. cFireSIM with a 50 year timeline shows similar results, except that the best results are with 100% equities and adding any amount of bonds or gold increase failure rates.

Bonds were quite valuable in 2000, and I think they probably are valuable for the psychological effects of mitigating volatility, but they don't actually improve historical results.

My conclusion is that chasing backtesting performance is pretty useless, because the sample size is too small and you end up overfitting unless your portfolio is very simple. So I trust a 4% guideline for some combination of 100/0 to 60/40 stocks/bonds and adding some gold or whatever probably doesn't move the needle much. Asset allocation matters less than a lower withdrawal rate. If 4% feels too risky then drop down to 3.5% or 3%. And don't trust anybody who claims that adding the correct blend of diversity changes results by a huge margin. That is, SCV or gold or REITs or whatever is not going to produce a reliable 5% withdrawal rate.

Better yet, pay attention after your income disappears. If you're 5 or 10 years into retirement and your portfolio is down by 30+% go generate some income.

aloevera1 · « **Reply #2 on:** October 12, 2024, 11:19:04 AM »

Agree, I see very limited use for them. One could use the term GIC's for storing cash in a ladder and to reduce volatility that way.

Bonds also don't really help with inflation much.

Scandium · « **Reply #3 on:** October 14, 2024, 02:16:33 PM »

Very interesting. I'll have to look more into the data as I get closer, but I think there may be some value to bonds to mitigate SOR? Going to 20-40% bonds a few years before/after retirement, then going, back to 100% stocks after that? At least that's what I got from ERN's analysis.

What is the timeline for gold prices? Could this largely be because gold price quadrupled from 2000 - 2011? Other than that the price was flat from 1981 - 2004..

SeattleCPA · « **Reply #4 on:** October 15, 2024, 07:03:38 AM »

Quote from: vand on October 12, 2024, 07:46:00 AM

But what I really what to do here is drill down here in just how terrible bonds are as a diversifier for FI purposes, and how poor they are at improving safe withdrawal rates.

On PortfolioCharts I took a portfolio of a 100% US Large Cap Blend, which by itself offers a SWR of 3.8% over 30yr as a baseline.
I then added in the following asset in 5% increments to see what effect it had on the SWR:

Short Term US Treasuries (1-3Y)
Intermediate Term US Treasuries (3-10Y)
Long Term US Treasuries (10-30Y)
US REITs
Gold
Commodities

<snip>

So I disagree, which won't surprise some people with good memory. I think bonds scrunch the range of outcomes. And that can be good. You give up upside and dodge some downside. I discussed this in a blog post here, Monte Carlo Simulations Show How Bonds Dampen Investment Risk, but you can understand their effect with a picture (see below). The red lines show all the outcomes with a 100% stocks portfolio the black lines show the worst and best case outcomes if you add treasuries.:

Regarding Tyler's PortfolioCharts, that is an amazing website. But the sample size is too small. You're essentially concluding you can use the period of history since, what, 1972 to predict future market returns?

SeattleCPA · « **Reply #5 on:** October 15, 2024, 07:08:34 AM »

Quote from: VanillaGorilla on October 12, 2024, 10:40:49 AM

I've done the same analysis many times and always come to the same conclusion: a 52 year timeframe is too small to be useful. The results are overfit to one or two cohorts: 1973 and maybe 2000.

I don't disagree with your conclusion that bonds are not very useful, historically. cFireSIM with a 50 year timeline shows similar results, except that the best results are with 100% equities and adding any amount of bonds or gold increase failure rates.

Bonds were quite valuable in 2000, and I think they probably are valuable for the psychological effects of mitigating volatility, but they don't actually improve historical results.

My conclusion is that chasing backtesting performance is pretty useless, because the sample size is too small and you end up overfitting unless your portfolio is very simple. So I trust a 4% guideline for some combination of 100/0 to 60/40 stocks/bonds and adding some gold or whatever probably doesn't move the needle much. Asset allocation matters less than a lower withdrawal rate. If 4% feels too risky then drop down to 3.5% or 3%. And don't trust anybody who claims that adding the correct blend of diversity changes results by a huge margin. That is, SCV or gold or REITs or whatever is not going to produce a reliable 5% withdrawal rate.

Better yet, pay attention after your income disappears. If you're 5 or 10 years into retirement and your portfolio is down by 30+% go generate some income.

Totally agree with above.

Again, I love PortfolioCharts. But I think for any portfolio you can backtest using a longer dataset, the PortfolioCharts numbers look too high. Thus, it seems to me you're really just assuming future mirrors recent past?

ChpBstrd · « **Reply #6 on:** October 15, 2024, 08:04:33 AM »

I'm chasing the same scent as @Scandium and wondering if these results have something to do with the epic multiplication in gold prices over the past 25 years - or the real estate manias of the 2000's and 2020's. I.e. how many simulated portfolios started between 50 and 30 years ago and were trending toward zero until they were rescued by run-ups in these specific assets over the past 20 years?

Second, I wonder if it can continue.

Can real estate continue generating more alpha than the broader stock market, and with less volatility, without attracting tons of investment money to extinguish that advantage? Arguably, the global post-pandemic run-up in real estate prices has already deeply reduced the future returns of this asset class. Also, arguably, global real estate subsidies like Freddie and Fannie Mae in the U.S. will eventually have to be scaled back when increasingly indebted governments can no longer afford them.

Can gold continue delivering stock-market like returns and behaving counter-cyclically to stocks? Or was the Chinese/Russian/Indian buying spree a one-time event, prior to some future decline of these economies? I don't think crypto will take the place of gold, but I could always be wrong about that.

Third, I wonder what would happen if one ran this analysis with corporate bonds? There is a wide range of risk tolerance between treasuries and REITs.

Fourth, when I step back and look, there seems to be something wrong with the chart (aside from the column labels). I've never heard anyone say the optimum stock/treasury AA is 50/50, but that's what the AA distribution shows. From every other source (e.g. ERN) I've read the highest SWR is somewhere closer to 80/20 or 90/10. What is different about this analysis that causes it to contradict all previous backtested advice I've read? The timeframe?

Fifth, are bonds really a bad diversifier? Do the data support that conclusion? Looks to me like they raised the SWR from 3.8% to 4.2%. That's not bad at all actually. It would mean one can retire with 10% less using this AA instead of an all-stock AA. For a household with an $80k spend rate, for example, that's the difference between having to earn an extra $200k prior to retirement! This actually tempts me to buy treasuries, which currently offer positive real yields and are theoretically an even better deal than in the past when they provided a much higher SWR.

Did other AAs work better? Yes, but there is a cherry-picking risk the more specific we get. We could say a portfolio of 1% large cap stock indices plus 99% NVDA would have been better, or 99% large cap indices plus 1% Bitcoin. But in terms of making forecasts we can be confident about, such obvious performance-chasing seems like a bad idea. Maybe gold and real estate have enjoyed big gains in recent decades of falling interest rates, and that's why we're looking at them instead of corporate/muni bonds, annuities, MLP's, agricultural futures, silver, mutual funds, or lots of other things that were on the investment menu 30-50 years ago?

ChpBstrd · « **Reply #7 on:** October 15, 2024, 08:30:32 AM »

I have done very well with various individual bond speculations over the years, and might be a better bond investor than stock investor. However, if I'm not locking in yields higher than my withdraw rate and actually retiring on the income, then my rationale for owning bonds is probably that a bear market is imminent, and that I'll trade them for stocks after the crash. To play that bond strategy, one would have to be a successful stock market timer! The people who merely rebalance are doing the same thing in a less dramatic manner, living more off the bonds when the stocks are down.

So instead of playing that game or permanently accepting the dirt-low yields treasuries offer, I use options to hedge my portfolio.

Diversification into bonds generally means accepting the probability of a lower return than the long-run average of stocks. However, I can use a collar strategy for free, nearly free, or for a small upfront credit.

A portfolio hedged this way behaves a lot like a 50/50 portfolio in terms of volatility, but is better because there is a firm floor on potential losses. Plus, growth is enhanced because I don't have a large chunk of my assets stranded in low-yielding or no-yielding assets like treasuries or gold. I can apply 100% of my portfolio to stocks and yet skip the historically typical 20-40% downturns in exchange for giving away any massive upturns. That's a deal I'll take in times of historically high stock valuation.

Finally, diversification assumes there will be counter-correlation between assets. E.g. bonds will do fine when stocks go down. That's a big assumption, as we learned in 2022 when both stocks and bonds fell hard together in response to rising interest rates. What's to say the same couldn't happen to gold or REITs? The backtests tell a story of at least some counter-correlation in the past, but we cannot say we know it will happen that way in the future. Suppose the same global crisis that causes stocks to sink also causes a central bank to dump gold? Or suppose another outbreak of high interest rates sinks both stocks and real estate together? Why are we relying on such guesswork?

My options, OTOH, are guaranteed by the exchanges, market makers, and participating brokerages. They are mathematically certain to counter-correlate my stocks, within boundary ranges I can foresee and choose. The collar strategy negates the time decay of options, and my hedge does not face the risk of damage if interest rates rise.

Thanks to LEAPS options markets for ETFs like SPY and QQQ, I can hedge for long-enough blocks of time - like 1.5 to 2 years. As my options approach expiration, I roll into the next duration on a low-VIX day 1-4 months ahead of expiration. This is generally less trouble than rebalancing.

As if setting my upside and downside outcomes and level of counter-correlation with mathematical certainty didn't offer enough control, the ability to time my rolls to occur on low-VIX days is the icing on the cake. What's not to like about collars, and what is to like about asset class diversification?

SeattleCPA · « **Reply #8 on:** October 15, 2024, 08:42:13 AM »

Quote from: ChpBstrd on October 15, 2024, 08:04:33 AM

Can real estate continue generating more alpha than the broader stock market, and with less volatility, without attracting tons of investment money to extinguish that advantage?

That Rate of Return on Everything paper I've often pointed to suggests it might continue to generate higher returns with lower volatility and show lack of correlation. (Link: https://academic.oup.com/qje/article/134/3/1225/5435538 )

Quote

Can gold continue delivering stock-market like returns and behaving counter-cyclically to stocks? Or was the Chinese/Russian/Indian buying spree a one-time event, prior to some future decline of these economies? I don't think crypto will take the place of gold, but I could always be wrong about that.

Pretty I've used cFireSim and PortfolioCharts to calculate impact of gold on asset allocations you can test. And cFireSim, looking at a much longer dataset, returns noticeably lower numbers... so I'd say your concern is extremely valid.

Scandium · « **Reply #9 on:** October 15, 2024, 08:44:17 AM »

Quote from: ChpBstrd on October 15, 2024, 08:04:33 AM

I'm chasing the same scent as @Scandium and wondering if these results have something to do with the epic multiplication in gold prices over the past 25 years - or the real estate manias of the 2000's and 2020's. I.e. how many simulated portfolios started between 50 and 30 years ago and were trending toward zero until they were rescued by run-ups in these specific assets over the past 20 years?

Second, I wonder if it can continue.

Can real estate continue generating more alpha than the broader stock market, and with less volatility, without attracting tons of investment money to extinguish that advantage? Arguably, the global post-pandemic run-up in real estate prices has already deeply reduced the future returns of this asset class. Also, arguably, global real estate subsidies like Freddie and Fannie Mae in the U.S. will eventually have to be scaled back when increasingly indebted governments can no longer afford them.

Thanks for validating my skepticism, lol. The more I look at this more questions I have actually.

I don't understand the REIT numbers. 60% REIT does better than 100% stocks?? A whole percentage point higher WR than 50-50 stocks-treasuries?! I find that hard to believe just on gut feel.. What index is this based on, and what is the timeframe? The VGSIX fund only goes back to 1996, hardly long enough.
https://www.morningstar.com/funds/xnas/vgsix/chart
It kept level while stocks crashed in ~2000, and outperformed in the runup to 2008, but other than that has done worse. So is this performance based on two very specific, short timeframes and confluence of events? I would not bet on that repeating. As was discussed in the REIT thread, IMO the bet on (continued increase of) income for ~150 commercial real estate landlords is not something I'm confident it. At the end of the day they are still stocks. The "fundamentals" of that business is not as sure in the future as loans to the US government.

Like you I'm actually very encouraged seeing that adding 30% treasuries bump the SWR from 3.8% to 4.2%! That's a decent increase, smooth returns, and should reduce the SOR as well from what I've seen from ERN. With minimal impact on long-term returns. Based on my current reading I will likely go to something around 20-30% bonds in the 5 years before/after retirement, and nice to see this validated.

vand · « **Reply #10 on:** October 15, 2024, 09:08:09 AM »

Quote from: Scandium on October 14, 2024, 02:16:33 PM

Very interesting. I'll have to look more into the data as I get closer, but I think there may be some value to bonds to mitigate SOR? Going to 20-40% bonds a few years before/after retirement, then going, back to 100% stocks after that? At least that's what I got from ERN's analysis.

What is the timeline for gold prices? Could this largely be because gold price quadrupled from 2000 - 2011? Other than that the price was flat from 1981 - 2004..

The way to think about the allocation of gold - or any component, is that the portfolio offered what it did at the very worst point for that asset. So a 65/35 stock/gold portfolio successfully supported a 5.3% 30yr SWR even starting from 1981 when gold was just entering its prolonged bear market. Remember. we're testing for the weakest points in any strategy or allocation decision.. not the average.

I'd actually like to get ERN's analysis on this too. I know he's considered adding gold to portfolios with the conclusion that it does improve SWRs slightly, but it was still from the baseline of "gold as the 3rd option". What if he just removes the bond component and works from first principals, using gold as the go-to diversifer?
https://earlyretirementnow.com/2020/01/08/gold-hedge-against-sequence-risk-swr-series-part-34/

Another important point to mention pertaining to Bonds is what Pensioncraft's Ramin asked about on the interview with Tyler -- that the PortfolioCharts Bond data is all unhedged, while the bond funds we see offered by Vanguard etc are nearly always currency hedged - this removes the effect of FX movements on the bond fund's performance, and it's likely that it will have some effect compared to an unhedged bond fund, but we're not exactly sure what the effect will be.. yes, I suppose it's possible that that 4.1-4.2% SWR for stock/bond portfolios could bump up a little if those bonds were hedged.. but without the relevant data series we can't know for sure.

vand · « **Reply #11 on:** October 15, 2024, 09:29:25 AM »

Quote from: SeattleCPA on October 15, 2024, 08:42:13 AM

Quote from: ChpBstrd on October 15, 2024, 08:04:33 AM
Can real estate continue generating more alpha than the broader stock market, and with less volatility, without attracting tons of investment money to extinguish that advantage?

That Rate of Return on Everything paper I've often pointed to suggests it might continue to generate higher returns with lower volatility and show lack of correlation. (Link: https://academic.oup.com/qje/article/134/3/1225/5435538 )

Quote
Can gold continue delivering stock-market like returns and behaving counter-cyclically to stocks? Or was the Chinese/Russian/Indian buying spree a one-time event, prior to some future decline of these economies? I don't think crypto will take the place of gold, but I could always be wrong about that.

Pretty I've used cFireSim and PortfolioCharts to calculate impact of gold on asset allocations you can test. And cFireSim, looking at a much longer dataset, returns noticeably lower numbers... so I'd say your concern is extremely valid.

I love CFireSim, but I'm pretty sure their treatment of Gold pricing pre-1972 is to just fix it at $35 (or the $20 fixed price prior to that), which is worse than useless and doesn't recognise the change in monetary system that occurred. Trying to backdate gold in your portfolio pre-1972 is a meaningless exercise..

Instead, it is worth considered what eventually caused Bretton Woods to fall apart, which was too much issuance of USDs which without the gold (and SDRs) to back them... it's not unreasonable that gold would have reacted well to these instabilities had it been allowed to float freely at that time, however we'll never really know.

Radagast · « **Reply #12 on:** October 15, 2024, 09:50:44 AM »

OK, sigh... we've been over this before on this forum but...

Safe withdrawal rates by definition do not include much data, and additional data doesn't usually help. They show the single worst year, and that's it. In this case that year is 1970, which is the single data point used in Portfolio Charts for the SWR for all countries (as far as I know), except Japan (also note all countries in the data set are France or border France, with 3 exceptions). Gold was uninvestable in the US in 1970, and began the year with its price fixed at an artificially low level which was then allowed to float on the market (not to the public) sometime during the year. So you cannot use Portfolio Charts to make any inferences with respect to gold and safe withdrawal rates, though it should be safe for backtesting with gold that is not with respect to withdrawal rates. Emphasis on the underlined part. Japan is the only independent data point in the Portfolio Charts data with a SWR determined in 1990, and coincidentally if you only use Japan you find the optimal allocation is 60% bonds.

Another critique I have of Portfolio Charts is that all of their back tests are for pessimistic outlooks, so it makes no sense to plot them against each other. Plotting minimum safe withdrawal rate against ulcer index doesn't makes sense because the market motions which result in a high ulcer index also result in low SWR. They're mostly the same thing. You could plot ulcer index against median return, or SWR against median return, which would both be conservative approaches, but it makes no sense to plot them against each other.

That said, I do think 10% gold, 10% short term bonds, and 10% long term bonds with the rest internationally diversified stocks and any portion of real estate is a near optimal allocation.

Also, there is plenty of evidence that diversification works, or at least it has worked very well in the past. There's not much evidence of any precise allocation that works so much as diversification in general works.

markbike528CBX · « **Reply #13 on:** October 15, 2024, 10:22:27 AM »

At least as early as 2008 James Grant described bonds as “Return-Free Risk”

https://advisoranalyst.com/2008/12/07/return-free-risk.html/

Also, Buffett

https://www.barrons.com/articles/warren-buffett-berkshire-hathaway-stocks-bonds-8474db0a

vand · « **Reply #14 on:** October 15, 2024, 10:25:04 AM »

Quote from: Radagast on October 15, 2024, 09:50:44 AM

OK, sigh... we've been over this before on this forum but...

Safe withdrawal rates by definition do not include much data, and additional data doesn't usually help. They show the single worst year, and that's it. In this case that year is 1970, which is the single data point used in Portfolio Charts for the SWR for all countries (as far as I know), except Japan (also note all countries in the data set are France or border France, with 3 exceptions). Gold was uninvestable in the US in 1970, and began the year with its price fixed at an artificially low level which was then allowed to float on the market (not to the public) sometime during the year. So you cannot use Portfolio Charts to make any inferences with respect to gold and safe withdrawal rates, though it should be safe for backtesting with gold that is not with respect to withdrawal rates. Emphasis on the underlined part. Japan is the only independent data point in the Portfolio Charts data with a SWR determined in 1990, and coincidentally if you only use Japan you find the optimal allocation is 60% bonds.

Another critique I have of Portfolio Charts is that all of their back tests are for pessimistic outlooks, so it makes no sense to plot them against each other. Plotting minimum safe withdrawal rate against ulcer index doesn't makes sense because the market motions which result in a high ulcer index also result in low SWR. They're mostly the same thing. You could plot ulcer index against median return, or SWR against median return, which would both be conservative approaches, but it makes no sense to plot them against each other.

That said, I do think 10% gold, 10% short term bonds, and 10% long term bonds with the rest internationally diversified stocks and any portion of real estate is a near optimal allocation.

Also, there is plenty of evidence that diversification works, or at least it has worked very well in the past. There's not much evidence of any precise allocation that works so much as diversification in general works.

Sigh... Let me repeat the SWR is determined from the weakest point, not the strongest. So in the case of gold, the supressed prices from 1971 (and the 3-year shadow market before that) isn't particularly relevant. The key date for stress testing a gold allocation is 1981, the peak of the gold market.

The data is what it is: on the PortfolioCharts global withdrawal rates simulations page you can remove Japan, and the worst scenarios by country filter are:

global: Japan 1990
excluding Japan: Australia 1987
exluding Japan & Australia: Spain 1970
excluding Japan, Australia & Spain: Italy 1974...

so clearly all road to failure don't simply lead to back to 1970 as you claim...

Radagast · « **Reply #15 on:** October 15, 2024, 11:08:01 AM »

Yup, but it doesn't matter. If it wasn't for the 1970 start, the optimal allocation to gold would backtest as much smaller, so far as I know for all countries. Limiting the test to when the public could actually invest in gold after 1975 would give completely different gold allocations, and I don't have the data to test this, but likely in the 0-20% range. Posters are speculating it was something to do with 2011 or whatever, but that's wrong. It's all determined by 1970, has been since 1970, and will be until the US has a worse set of market conditions than existed in 1970. Ditto for REITs and all investments for that that matter in the US. 1970 was the worst case, so the optimal allocation for a high SWR was determined by what worked best for the cohort starting in 1970 which didn't sand bag any other cohorts. However, at least other allocations were more or less investable in a similar manner as they are today.

ChpBstrd · « **Reply #16 on:** October 15, 2024, 11:20:15 AM »

The first REIT to go public (Kim Realty) didn't do so until 1991, so I'd argue there are only about three historical 30-year cohorts to evaluate for that asset class: part of 1991, 1992, and 1993. 1994's results will be available in 2.5 months.

Radagast · « **Reply #17 on:** October 15, 2024, 11:54:34 AM »

That is true about REITs, though I don't have a reason to think an average investor in shopping malls, office space, or apartments in 1970 would have done significantly different than the data for REITs (in fact they might have done better). However, I can see that what gold did in 1970 was entirely artificial and has no semblance to its subsequent pricing or behaviour.

Radagast · « **Reply #18 on:** October 15, 2024, 12:58:27 PM »

Thought I'd also add this since it seems to have passed out of popular discourse: Portfolio Charts data is taken from the Simba Spreadsheet which is maintained at Bogleheads. So far as I know, it is the most extensive and thoroughly validated open source of asset class returns. You can download the spreadsheet and use the data as you will:

https://www.bogleheads.org/wiki/Simba%27s_backtesting_spreadsheet

SeattleCPA · « **Reply #19 on:** October 16, 2024, 06:38:04 AM »

Quote from: Radagast on October 15, 2024, 12:58:27 PM

Thought I'd also add this since it seems to have passed out of popular discourse: Portfolio Charts data is taken from the Simba Spreadsheet which is maintained at Bogleheads. So far as I know, it is the most extensive and thoroughly validated open source of asset class returns. You can download the spreadsheet and use the data as you will:

https://www.bogleheads.org/wiki/Simba%27s_backtesting_spreadsheet

The Simba spreadsheet, at least per Tyler's website, is just one of the data sources. (Cite: https://portfoliocharts.com/user-guide/data-sources/ )

Regarding cFirmSim not providing gold price or acceptable gold price info, that's possible. (Sound like it's using the fixed US price?) So this idea in terms of using PortfolioCharts.com to estimate SWRs for various asset allocations: Try some other allocation in both PortfolioCharts and cFireSim where you can compare the numbers from the shorter dataset to the longer dataset.

I think the shorter dataset seems to always be more optimistic. And I guess my main point is, the sample size is usually too small. (I am not saying Tyler's website is bad. It isn't. It's excellent. I love it.)

The problem with these historical lookbacks? Even if you work from datasets that start in 1872? You've basically got two truly distinct samples if you're talking about say 30 years of work and 30 years of retirement. There's the six decade stretch that starts in 1872 and runs through 1932. And then the one that starts in 1932 and runs through 1992.

We won't know about the six decade stretch that started in 1992 until 2052?

ChpBstrd · « **Reply #20 on:** October 16, 2024, 07:26:44 AM »

Quote from: SeattleCPA on October 16, 2024, 06:38:04 AM

I think the shorter dataset seems to always be more optimistic.

I think we also over-estimate the availability / normality of certain investments in an age when there's an ETF for everything and we click illuminated parts of glass to instantaneously enter and exit asset classes. We forget how previous generations had to set appointments to meet with brokers or other sellers and negotiate the purchase of everything.

Gold was illegal to invest in, and impossible for regular people to buy in bulk, for the vast majority of the 20th century. I do recall my grandfather buying a heavy brick of silver from Morgan Stanley in the mid 1980s, and wheeling it around the house on a dolly. He later said he lost money on it. But that wasn't during the worst of the inflation in the 1970s, so why are we pretending a person could have clicked the GLD or IAU button back then? Being an individual buying a gold bar in 1971 would have been as weird as keeping a storage unit with a couple pallets of copper or lead ingots today. And if you don't factor in the cost of a substantial safe, we need to discount returns for the risk of theft - something nobody even thinks about with funds.

And as noted the first REIT wasn't publicly traded until '91, so your only option as a regular Joe would have been to find other investors looking for passive partners, and that is a recipe for getting ripped off by shady characters if I've ever heard one. You could have taken an active role managing rent houses or storefronts, but the 70s and 80s were also a period of urban decline so the places where you'd have been most likely to buy might have suffered years of negative real appreciation. Maybe the developers of malls and housing developments did OK, but those investments were generally not available to individuals with a couple thousand bucks to allocate.

Overall everything was more granular, and one's returns depended completely on the specifics of each deal you could negotiate in person with another human being. I don't think any investment return series fully captures the variance and risk one had to take in order to diversify in the past. They certainly don't capture the effects of mutual fund loads or $1,500 commissions.

Electronic markets and the rise of low-cost funds democratized investing starting in the late 1990s, and arguably enabled more money to flow into certain assets. Never before had small investors been able to click in and out of things like various precious metals, oil or gas futures, funds containing hundreds of stocks, entire diversified market sectors, equal weighted index funds, leveraged funds, inverse funds, and so on. Until the 21st century, such tools were still a novelty for retirees, and most money was in mutual funds bought through live brokers for substantial commissions.

So in terms of what was actually on the menu, most passive investors from the 60's through the 90's could realistically choose from the following, and to do otherwise was weird and prohibitively expensive:

Bank CDs or savings accounts
Mutual funds purchased through a broker with large upfront loads and commissions
Individual corporate bonds purchased through a broker " " " " " "
Individual stocks purchased through a broker " " " " " "
Treasuries or savings bonds
Physically held foreign currency notes
For higher net worth investors, oil field royalties or partnerships, or passive partnerships in small private companies
Undeveloped land, held for its appreciation potential at the cost of property taxes and opportunity cost

Also, of course, paying off one's mortgage made sense in this era, as it was by far the highest-yielding risk free investment. Most people never invested beyond doing this.

vand · « **Reply #21 on:** October 16, 2024, 09:17:29 AM »

Quote from: ChpBstrd on October 16, 2024, 07:26:44 AM

Quote from: SeattleCPA on October 16, 2024, 06:38:04 AM
I think the shorter dataset seems to always be more optimistic.
I think we also over-estimate the availability / normality of certain investments in an age when there's an ETF for everything and we click illuminated parts of glass to instantaneously enter and exit asset classes. We forget how previous generations had to set appointments to meet with brokers or other sellers and negotiate the purchase of everything.

Gold was illegal to invest in, and impossible for regular people to buy in bulk, for the vast majority of the 20th century. I do recall my grandfather buying a heavy brick of silver from Morgan Stanley in the mid 1980s, and wheeling it around the house on a dolly. He later said he lost money on it. But that wasn't during the worst of the inflation in the 1970s, so why are we pretending a person could have clicked the GLD or IAU button back then? Being an individual buying a gold bar in 1971 would have been as weird as keeping a storage unit with a couple pallets of copper or lead ingots today. And if you don't factor in the cost of a substantial safe, we need to discount returns for the risk of theft - something nobody even thinks about with funds.

And as noted the first REIT wasn't publicly traded until '91, so your only option as a regular Joe would have been to find other investors looking for passive partners, and that is a recipe for getting ripped off by shady characters if I've ever heard one. You could have taken an active role managing rent houses or storefronts, but the 70s and 80s were also a period of urban decline so the places where you'd have been most likely to buy might have suffered years of negative real appreciation. Maybe the developers of malls and housing developments did OK, but those investments were generally not available to individuals with a couple thousand bucks to allocate.

Overall everything was more granular, and one's returns depended completely on the specifics of each deal you could negotiate in person with another human being. I don't think any investment return series fully captures the variance and risk one had to take in order to diversify in the past. They certainly don't capture the effects of mutual fund loads or $1,500 commissions.

Electronic markets and the rise of low-cost funds democratized investing starting in the late 1990s, and arguably enabled more money to flow into certain assets. Never before had small investors been able to click in and out of things like various precious metals, oil or gas futures, funds containing hundreds of stocks, entire diversified market sectors, equal weighted index funds, leveraged funds, inverse funds, and so on. Until the 21st century, such tools were still a novelty for retirees, and most money was in mutual funds bought through live brokers for substantial commissions.

So in terms of what was actually on the menu, most passive investors from the 60's through the 90's could realistically choose from the following, and to do otherwise was weird and prohibitively expensive:
Bank CDs or savings accounts
Mutual funds purchased through a broker with large upfront loads and commissions
Individual corporate bonds purchased through a broker " " " " " "
Individual stocks purchased through a broker " " " " " "
Treasuries or savings bonds
Physically held foreign currency notes
For higher net worth investors, oil field royalties or partnerships, or passive partnerships in small private companies
Undeveloped land, held for its appreciation potential at the cost of property taxes and opportunity cost

Also, of course, paying off one's mortgage made sense in this era, as it was by far the highest-yielding risk free investment. Most people never invested beyond doing this.

Yes, agree 100%. The financialisation of markets has increased the ease and and accessibility of investing into a seemingly limitless range of asset classes and investing themes that simply weren't possible for previous generations - and at such low cost, too. Today's investors have never had it so good in this regard, and previous generations were far more limited to what they could realistically access. However, as a counterpoint, the concept of all-weather portfolios using gold has been around at least as far back as 1982 when Harry Browne was first floating the idea of the Permanent Portfolio, so it's not new, and certainly before the time when IGU or GLD was a thing.

Incidentally, a side of effct of the financialisation/accessibility of markets is probably that correlation has risen between most asset classes due to this accessibility - if everyone's has access to the same markets its reasonable to think that the waves of fear and greed seep from one holding to another in an investor's portfolio.

regarding PortfolioCharts' data range, I can understand why someone would place more faith in a backtest that has more sample years (wouldn't we all).. maybe in an parallel universe the 1933-71 gold fix would never have been a thing and we'd be able to compare apples to apples, but that's just the hand we're dealt when analysing this stuff. However, the PortfolioCharts numbers are still very fair in the sense that the starting point is the same for all asset classes considered (1970) - which eliminates the depression (bad for stocks) and anything the late 1960s (bad for bond/stocks) as potential start dates, and yet even since just 1970 all combinations of stock/bonds still can't manage better than 4.2% SWR.

vand · « **Reply #22 on:** October 17, 2024, 04:38:08 AM »

I am sure I've posted these video links before from belangp, but I'd encourage people to watch them here:

Stock/Bond/Gold Mixes
Even this video done in 2020 suggested that a 65/35 stock/gold mix is probably near the optimum in terms of portfolio efficiency, with no real need bonds. This completely reaffirms my findings above where I found that a 65/35 stock/gold allocation also provided the highest SWR for all stock/gold/bond combinations

https://www.youtube.com/watch?v=LA2Yr6NoZyA
(note that for return numbers he's used the trendline return, not the actual return, so smoothing out for the gold's depressed starting price in 1972 which would unfairly flatter it)

The key chart:

The Yellow line is all combinations of Stock/Gold
The Red line is all combinations of Stocks/Bonds

The Orange area represent portfolios combinations that are better than their stock/bond counterparts (keeping the stock portion constant) by swapping some or all of the bond allocation to gold.

Why Monte Carlo is deeply flawed in our assumed understanding of failure rates (and why it flatters stock/bond mixes):
https://www.youtube.com/watch?v=DF7nrI0XWFY

MustacheAndaHalf · « **Reply #23 on:** October 20, 2024, 11:26:29 PM »

Notice the start date: 1972. People pushing gold ALWAYS include 1972-1974, which were 3 of the best 4 years of gold's performance. What gets left out is U.S. leaving the gold standard in 1971, which ended the fixed price of gold. The U.S. won't leave the gold standard twice, so performance from 1972-1974 will not repeat.

1972 - 2024: gold returned 8%/year
1975 - 2024: gold returned 5%/year

(According to Portfolio Visualizer, using portfolio of 100% gold)
https://www.portfoliovisualizer.com/backtest-asset-class-allocation

Dicey · « **Reply #24 on:** October 21, 2024, 01:57:04 AM »

I live in a HCOLA, and pre-FIRE always had a relatively huge mortgage. In hindsight, I wish I had considered my mortgage part of my bond allocation, and invested more heavily in equities. A 60/40, or even 70/30 split, plus a big-ass fixed rate mortgage is way too conservative.

SilentC · « **Reply #25 on:** October 21, 2024, 08:27:37 PM »

Quote from: Dicey on October 21, 2024, 01:57:04 AM

I live in a HCOLA, and pre-FIRE always had a relatively huge mortgage. In hindsight, I wish I had considered my mortgage part of my bond allocation, and invested more heavily in equities. A 60/40, or even 70/30 split, plus a big-ass fixed rate mortgage is way too conservative.

I think you probably did fine, if you have a large mortgage that’s like the opposite of having too much fixed income allocation (you are borrowing so you are short fixed income/leveraged), so you were probably not being that conservative. But yeah owning more stocks the last 14 years was an epic trade.

Radagast · « **Reply #26 on:** October 21, 2024, 08:45:20 PM »

My own experiments with Portfolio Visualizer (no longer useable) did not show nearly as much gold being optimum. Granted this was for periods starting 1992 through 2006, depending on how picky I wanted to be using actual relevant funds. Bonds did great during these periods, with strong negative correlations exactly when you wanted them.

Bill Bernstein in The Intelligent Asset Allocator showed that the best efficient frontier for one period was generally markedly different from subsequent period. I expect that this is again an issue of picking a start date from 1970 to 1974 when it was literally illegal for an individual investor in the US to own gold, so these results are literally impossible for a real person.

vand · « **Reply #27 on:** October 22, 2024, 03:44:37 AM »

At the risk of feeding the troll...

Quote from: MustacheAndaHalf on October 20, 2024, 11:26:29 PM

Notice the start date: 1972. People pushing gold ALWAYS include 1972-1974, which were 3 of the best 4 years of gold's performance. What gets left out is U.S. leaving the gold standard in 1971, which ended the fixed price of gold. The U.S. won't leave the gold standard twice, so performance from 1972-1974 will not repeat.

1972 - 2024: gold returned 8%/year
1975 - 2024: gold returned 5%/year

(According to Portfolio Visualizer, using portfolio of 100% gold)
https://www.portfoliovisualizer.com/backtest-asset-class-allocation

So what? what's that got to do with the >diversifying< qualities of gold?

As already stated more than once, it's not the >best< outcome that determines the SWR it's the >worst< outcome. A Stock/Gold portfolio was able to sustain a 5.3% withdrawal rate even if you had retired from the very worst time - for both gold by itself, or for the portfolio as a whole. This stuff is worth talking about because it IS more complicated than just looking at the absolute return of individual components from cherry picked years... ~~Some day you gotta advance past kindergarten!~~

<back on ignore>

MOD NOTE: Read rule #1 of the forum, please.

vand · « **Reply #28 on:** October 22, 2024, 03:48:36 AM »

Quote from: Radagast on October 21, 2024, 08:45:20 PM

My own experiments with Portfolio Visualizer (no longer useable) did not show nearly as much gold being optimum. Granted this was for periods starting 1992 through 2006, depending on how picky I wanted to be using actual relevant funds. Bonds did great during these periods, with strong negative correlations exactly when you wanted them.

Bill Bernstein in The Intelligent Asset Allocator showed that the best efficient frontier for one period was generally markedly different from subsequent period. I expect that this is again an issue of picking a start date from 1970 to 1974 when it was literally illegal for an individual investor in the US to own gold, so these results are literally impossible for a real person.

PV is still very useable (moreso than ever), although they've just change the layout a bit so its not quite as intuitive. It's is a great tool but as you can only run a single simulation at a time the results you get are always going to be quite start-date sensitive.

One of the the problem with PV is that their datasets can be quite limited. For stocks they have US data back to 1972,but only non-US data from 1996. Similar for bonds. For other asset classes the data may be even more limited. In practice this makes it very US-centric in its view of outcomes, and for some fairly common mixes that you may want to test for you can't even get enough data for full single 30yr period.

Also, why would you only want to test starting year between 1992-2006? Surely if there's more data you want to use the full dataset available.

vand · « **Reply #29 on:** October 22, 2024, 04:00:58 AM »

Quote from: Dicey on October 21, 2024, 01:57:04 AM

I live in a HCOLA, and pre-FIRE always had a relatively huge mortgage. In hindsight, I wish I had considered my mortgage part of my bond allocation, and invested more heavily in equities. A 60/40, or even 70/30 split, plus a big-ass fixed rate mortgage is way too conservative.

Yes, an excellent point which we have made before on these pages, especially if you have a ARM.

It's wild to think that if you borrow 3-4 times your salary to buy a house you have all that interest rate exposure, then you'd go off and add even more to your investment portfolio. There's a disconnect there that gets missed by the majority of even traditional financial advice.

SilentC · « **Reply #30 on:** October 22, 2024, 06:03:45 AM »

Quote from: vand on October 22, 2024, 04:00:58 AM

Quote from: Dicey on October 21, 2024, 01:57:04 AM
I live in a HCOLA, and pre-FIRE always had a relatively huge mortgage. In hindsight, I wish I had considered my mortgage part of my bond allocation, and invested more heavily in equities. A 60/40, or even 70/30 split, plus a big-ass fixed rate mortgage is way too conservative.

Yes, an excellent point which we have made before on these pages, especially if you have a ARM.

It's wild to think that if you borrow 3-4 times your salary to buy a house you have all that interest rate exposure, then you'd go off and add even more to your investment portfolio. There's a disconnect there that gets missed by the majority of even traditional financial advice.

I replied to a prior post on this too, maybe I’m missing something? The way I see it, if you have a mortgage and a fixed income portfolio they oppose each other. Why are you considering having a mortgage and owning fixed income (which could be mortgages) equivalent? If rates go up due to inflation or currency devaluation you win as a mortgage holder as you have a below market interest rate you can pay back with cheaper dollars while you lose on your fixed income portfolio which decreases in value. If there is deflation or currency strengthening the opposite is true, though the mortgage may be more “hedged” against this with a built in refinancing option the mortgage holder pays for (which is rolled into the mortgage rate and today worth close to 0.8% per year).

Am I making this too hard or missing something? Thanks.

ChpBstrd · « **Reply #31 on:** October 22, 2024, 06:38:12 AM »

Quote from: SilentC on October 22, 2024, 06:03:45 AM

Quote from: vand on October 22, 2024, 04:00:58 AM
Quote from: Dicey on October 21, 2024, 01:57:04 AM
I live in a HCOLA, and pre-FIRE always had a relatively huge mortgage. In hindsight, I wish I had considered my mortgage part of my bond allocation, and invested more heavily in equities. A 60/40, or even 70/30 split, plus a big-ass fixed rate mortgage is way too conservative.
Yes, an excellent point which we have made before on these pages, especially if you have a ARM.

It's wild to think that if you borrow 3-4 times your salary to buy a house you have all that interest rate exposure, then you'd go off and add even more to your investment portfolio. There's a disconnect there that gets missed by the majority of even traditional financial advice.
I replied to a prior post on this too, maybe I’m missing something? The way I see it, if you have a mortgage and a fixed income portfolio they oppose each other. ...

Am I making this too hard or missing something? Thanks.

I can think of 2 possibilities aside from the recent phenomenon of few-year-old mortgages having lower interest rates than government bonds:

1) A person with mortgage+bonds arguably has more quickly accessible liquidity than a person with neither. This could be relevant to those of us in the cutthroat US where being able to come up with a six figure sum for medical procedures and drugs is often the difference between living and dying.

2) Suppose you are a person concerned about a real estate bubble who lives in a non-recourse state and who has concerns that the HCOLA house you bought with only 3% or 5% down could lose 20-30% of its value. If you POYM, you have full exposure to that risk of equity loss. If you hold the mortgage, you maintain the option to allow the house to be foreclosed in the event of a housing crash, thereby only losing a fraction of the equity. Then you have enough bonds to sell that you could make the next down payment at a much lower price. Seen in this light, a house with a mortgage functions like stock with a put option. It insulates recent homeowners from downside risk. A homebuyer might choose to configure themselves this way even if the bonds yielded less than the mortgage, just like stockholders who buy protective puts are willing to pay the cost of time decay in exchange for the protection. Note that non-recourse states include some of the hottest real estate markets, like California, Oregon, Connecticut, Arizona, Texas, and Washington. Perhaps they're hot markets because of this dynamic? In reality lenders in recourse states rarely find it worth the effort to shake extra cash out of the people who've foreclosed.

vand · « **Reply #32 on:** October 22, 2024, 08:07:21 AM »

Quote from: SilentC on October 22, 2024, 06:03:45 AM

I replied to a prior post on this too, maybe I’m missing something? The way I see it, if you have a mortgage and a fixed income portfolio they oppose each other. Why are you considering having a mortgage and owning fixed income (which could be mortgages) equivalent? If rates go up due to inflation or currency devaluation you win as a mortgage holder as you have a below market interest rate you can pay back with cheaper dollars while you lose on your fixed income portfolio which decreases in value. If there is deflation or currency strengthening the opposite is true, though the mortgage may be more “hedged” against this with a built in refinancing option the mortgage holder pays for (which is rolled into the mortgage rate and today worth close to 0.8% per year).

Am I making this too hard or missing something? Thanks.

Uh, probably just you making it too hard.

In an environment where rates are going up it means borrowing gets more expensive (ie your mortgage) and bond holders are seeing the value of those bonds fall (ie your investment portfolio). Double whammy.

(sure, I know in US its common to have a 30yr fixed mortgage, but that's not common elsewhere, and even in the US you have some people on ARMs)

SilentC · « **Reply #33 on:** October 22, 2024, 08:29:10 AM »

Quote from: vand on October 22, 2024, 08:07:21 AM

Quote from: SilentC on October 22, 2024, 06:03:45 AM

I replied to a prior post on this too, maybe I’m missing something? The way I see it, if you have a mortgage and a fixed income portfolio they oppose each other. Why are you considering having a mortgage and owning fixed income (which could be mortgages) equivalent? If rates go up due to inflation or currency devaluation you win as a mortgage holder as you have a below market interest rate you can pay back with cheaper dollars while you lose on your fixed income portfolio which decreases in value. If there is deflation or currency strengthening the opposite is true, though the mortgage may be more “hedged” against this with a built in refinancing option the mortgage holder pays for (which is rolled into the mortgage rate and today worth close to 0.8% per year).

Am I making this too hard or missing something? Thanks.

Uh, probably just you making it too hard.

In an environment where rates are going up it means borrowing gets more expensive (ie your mortgage) and bond holders are seeing the value of those bonds fall (ie your investment portfolio). Double whammy.

(sure, I know in US its common to have a 30yr fixed mortgage, but that's not common elsewhere, and even in the US you have some people on ARMs)

But you already have the mortgage. It’s not getting more expensive for you if rates increase, it’s starting to become very valuable to you and why people don’t want to move to keep that value. Not double whammy, right? Opposite of double whammy.

vand · « **Reply #34 on:** October 22, 2024, 09:17:40 AM »

Quote from: SilentC on October 22, 2024, 08:29:10 AM

Quote from: vand on October 22, 2024, 08:07:21 AM
Quote from: SilentC on October 22, 2024, 06:03:45 AM

I replied to a prior post on this too, maybe I’m missing something? The way I see it, if you have a mortgage and a fixed income portfolio they oppose each other. Why are you considering having a mortgage and owning fixed income (which could be mortgages) equivalent? If rates go up due to inflation or currency devaluation you win as a mortgage holder as you have a below market interest rate you can pay back with cheaper dollars while you lose on your fixed income portfolio which decreases in value. If there is deflation or currency strengthening the opposite is true, though the mortgage may be more “hedged” against this with a built in refinancing option the mortgage holder pays for (which is rolled into the mortgage rate and today worth close to 0.8% per year).

Am I making this too hard or missing something? Thanks.

Uh, probably just you making it too hard.

In an environment where rates are going up it means borrowing gets more expensive (ie your mortgage) and bond holders are seeing the value of those bonds fall (ie your investment portfolio). Double whammy.

(sure, I know in US its common to have a 30yr fixed mortgage, but that's not common elsewhere, and even in the US you have some people on ARMs)

But you already have the mortgage. It’s not getting more expensive for you if rates increase, it’s starting to become very valuable to you and why people don’t want to move to keep that value. Not double whammy, right? Opposite of double whammy.

It might get more expensive if you need to move and refinance at some point, or if, as I said, if you are one of the many people on an adjustable rate mortgage.

If you have a mortgage and rates go down then you still get a benefit through being able to finance to a cheaper rate. Remember the theme of the topic - diversification. It's about risk management and not putting all your eggs into any one particular future version of the world.

Radagast · « **Reply #35 on:** October 22, 2024, 10:08:07 AM »

Quote from: vand on October 22, 2024, 03:48:36 AM

Quote from: Radagast on October 21, 2024, 08:45:20 PM
My own experiments with Portfolio Visualizer (no longer useable) did not show nearly as much gold being optimum. Granted this was for periods starting 1992 through 2006, depending on how picky I wanted to be using actual relevant funds. Bonds did great during these periods, with strong negative correlations exactly when you wanted them.

Bill Bernstein in The Intelligent Asset Allocator showed that the best efficient frontier for one period was generally markedly different from subsequent period. I expect that this is again an issue of picking a start date from 1970 to 1974 when it was literally illegal for an individual investor in the US to own gold, so these results are literally impossible for a real person.

PV is still very useable (moreso than ever), although they've just change the layout a bit so its not quite as intuitive. It's is a great tool but as you can only run a single simulation at a time the results you get are always going to be quite start-date sensitive.

One of the the problem with PV is that their datasets can be quite limited. For stocks they have US data back to 1972,but only non-US data from 1996. Similar for bonds. For other asset classes the data may be even more limited. In practice this makes it very US-centric in its view of outcomes, and for some fairly common mixes that you may want to test for you can't even get enough data for full single 30yr period.

Also, why would you only want to test starting year between 1992-2006? Surely if there's more data you want to use the full dataset available.

I'd already done lots of testing using generic asset classes, so I wanted to use actual funds similar to what I can buy today whenever possible. In some cases I could make reasonable approximations back to the 1980's, for example I used half Vanguard active international growth and active international value to approximate international.

Radagast · « **Reply #36 on:** October 22, 2024, 10:13:23 AM »

Quote from: SilentC on October 22, 2024, 06:03:45 AM

Quote from: vand on October 22, 2024, 04:00:58 AM
Quote from: Dicey on October 21, 2024, 01:57:04 AM
I live in a HCOLA, and pre-FIRE always had a relatively huge mortgage. In hindsight, I wish I had considered my mortgage part of my bond allocation, and invested more heavily in equities. A 60/40, or even 70/30 split, plus a big-ass fixed rate mortgage is way too conservative.

Yes, an excellent point which we have made before on these pages, especially if you have a ARM.

It's wild to think that if you borrow 3-4 times your salary to buy a house you have all that interest rate exposure, then you'd go off and add even more to your investment portfolio. There's a disconnect there that gets missed by the majority of even traditional financial advice.

I replied to a prior post on this too, maybe I’m missing something? The way I see it, if you have a mortgage and a fixed income portfolio they oppose each other. Why are you considering having a mortgage and owning fixed income (which could be mortgages) equivalent? If rates go up due to inflation or currency devaluation you win as a mortgage holder as you have a below market interest rate you can pay back with cheaper dollars while you lose on your fixed income portfolio which decreases in value. If there is deflation or currency strengthening the opposite is true, though the mortgage may be more “hedged” against this with a built in refinancing option the mortgage holder pays for (which is rolled into the mortgage rate and today worth close to 0.8% per year).

Am I making this too hard or missing something? Thanks.

First, because all bond funds yield higher than my 2.75% mortgage right now.
Second, because bonds as part of an asset allocation cannot be viewed independently from that allocation. Their purpose is to improve the risk/return of the portfolio, and a mortgage is not a substitute for this. As an extreme but real example, in spring of 2020 I was selling ZROZ to buy RZV, which was heck of a trade, and I needed to own bonds to do it. Even though my mortgage rate was much higher than the bond yield.

SilentC · « **Reply #37 on:** October 22, 2024, 10:30:02 AM »

@Radagast and @vand I read your replies and I still don’t get it. I look at things from a corporate finance lens which may be the reason. @vand I also am thinking US, where 92% of mortgages are fixed.

How does being short a mortgage backed security through having a mortgage add to your fixed income allocation? That’s like me saying I have $100k of Verizon bonds in my portfolio and I shorted $100k of AT&T bonds and then claiming my net fixed income exposure is $200k, when it’s closer to $0 net.

If I have a $1mn net worth which is $200k in high yield bond, $1.3mn equities and a 500k margin loan or mortgage, do I say I have $700k of fixed income in my portfolio? I think that my asset allocation in this example is 87% equity 13% fixed income personally. How is owing someone $500k the same as someone owing me $500k? I get that there can be negative convexity to a mortgage which changes the equation (but not for @Radagast who has very positive convex mortgage) but big picture being levered isn’t the same as fixed income exposure in my thinking. Maybe someone can finance talk/math this for me?

Radagast · « **Reply #38 on:** October 22, 2024, 10:40:13 AM »

I don't know corporate finance, but my understanding is that corporations may have both cash reserves and debt, simultaneously. Different functions, right? And a bank may borrow money at one rate to loan it out at another right? So this extends to individuals too.

I consider my mortgage part of the value of my property, rather than my investments, and I subtract mortgage principal from estimated home valuation to determine how much house I own. Mental accounting, sure but that's what I do. It doesn't factor into my investment portfolio.

Radagast · « **Reply #39 on:** October 22, 2024, 11:20:36 AM »

Sorry SilentC I think misread your question. Yes, a mortgage should be considered a negative bond allocation. It's the opposite of a bond allocation. It's not fixed income, it's fixed expenditure. It is not in any way a substitute for a bond allocation.

ChpBstrd · « **Reply #40 on:** October 22, 2024, 12:33:41 PM »

The complexity of the mortgage-as-a-short-bond conversation illustrates why most people think of their house payment as a simple cash expense.

Differences between having a mortgage and owning a short bond:

With a short bond you have to pay the owner interest on your loan from them plus the interest the bond would have paid them, but with a mortgage you only pay interest to the lender to borrow their cash.
You cannot exit a mortgage for a capital gain or loss like you can with a short bond.
A mortgage cannot be called by your broker.
A mortgage is generally treated differently in events like divorce. In a forced refinancing, you can lose your low interest rate loan with no compensation if rates have risen.

SilentC · « **Reply #41 on:** October 22, 2024, 03:06:40 PM »

Quote from: ChpBstrd on October 22, 2024, 12:33:41 PM

The complexity of the mortgage-as-a-short-bond conversation illustrates why most people think of their house payment as a simple cash expense.

Differences between having a mortgage and owning a short bond:
With a short bond you have to pay the owner interest on your loan from them plus the interest the bond would have paid them, but with a mortgage you only pay interest to the lender to borrow their cash.
You cannot exit a mortgage for a capital gain or loss like you can with a short bond.
A mortgage cannot be called by your broker.
A mortgage is generally treated differently in events like divorce. In a forced refinancing, you can lose your low interest rate loan with no compensation if rates have risen.

Agreed @ChpBstrd and thanks @, it’s much more complex having a mortgage vs actually being short a bond. The point I was after was that I didn’t understand why one should consider a having a mortgage as a reason to own less fixed income, even though having leverage and owning fixed income are more or less opposites. But I have been proven conceptually wrong here before so I wanted to make sure there wasn’t a blind spot in that thinking.

ChpBstrd · « **Reply #42 on:** October 22, 2024, 03:58:10 PM »

Quote from: SilentC on October 22, 2024, 03:06:40 PM

Quote from: ChpBstrd on October 22, 2024, 12:33:41 PM
The complexity of the mortgage-as-a-short-bond conversation illustrates why most people think of their house payment as a simple cash expense.

Differences between having a mortgage and owning a short bond:
With a short bond you have to pay the owner interest on your loan from them plus the interest the bond would have paid them, but with a mortgage you only pay interest to the lender to borrow their cash.
You cannot exit a mortgage for a capital gain or loss like you can with a short bond.
A mortgage cannot be called by your broker.
A mortgage is generally treated differently in events like divorce. In a forced refinancing, you can lose your low interest rate loan with no compensation if rates have risen.

Agreed @ChpBstrd and thanks @, it’s much more complex having a mortgage vs actually being short a bond. The point I was after was that I didn’t understand why one should consider a having a mortgage as a reason to own less fixed income, even though having leverage and owning fixed income are more or less opposites. But I have been proven conceptually wrong here before so I wanted to make sure there wasn’t a blind spot in that thinking.

I think it's even trickier to ask oneself if you would take out a non-callable margin loan to buy stocks at the interest rate of your mortgage.

For me, with my 3.25% loan, the answer is yes, I'd take that bet. In fact I'm taking it now. It's a no-brainer as far as I'm concerned. But if I had a more recent mortgage in the 6.5% to 7% range I'd have a much harder time justifying the risks of leverage for a probably much smaller net reward.

Again, this is deviating from thinking of the house payment as a simple rent-like expense like most people do. But we have to justify to ourselves why we don't think of our mortgage as a margin loan regardless of what investments we've purchased on the side. Intuition tells us this is a consequential decision, not just a game of mental accounting. Those with mortgage rates in the mid-single-digits or very expensive houses have a tougher task than me.

A corporation would evaluate the expected return on investment/expansion ideas against their cost of capital from selling stock and bonds at their planned/optimized future capitalization structure. If a new facility is expected to earn 10% of its cost per year, and the corporation has a weighted average cost of capital of, say, 9%, then the project will be a tough sell because it is a risky way to earn 1% extra compared to the completely safe alternative of paying down debt and buying back stock. Given any reasonable range of error in the estimate, that project could easily turn out to be destructive to economic value.

I cannot fault a household "board of directors" who chooses to pay down their 6.5% mortgage rather than investing in stocks or bonds. 6.5% is their risk-free rate of return and their cost of capital. The risky rate of return on stocks - let's say 10% plus or minus 30% - must have the cost of capital deducted to arrive at an expected net return, and that expected net return might be judged by reasonable people to be too small to justify taking the risk of a very bad scenario, such as suffering a bear market while paying interest!

On a deeper level, a household's cost of capital is their highest-interest-rate debt outstanding, or the lowest-interest-rate debt they could take on. So if a household was carrying credit card debt at 25%, there is NO WAY they should own any investment until that is paid off. Similarly, if a household can get a HELOC at x%, that is essentially their margin rate. This should all make sense on an intuitive level.

TL;DR: I say no to thinking about a mortgage as a negative bond, but yes to thinking about it as a household's cost of capital or margin loan rate. This is how (competent) corporate leaders make decisions, we can incorporate it into what else we know about financial theory, and there are fewer contradictions than with the negative-bond approach.

MustacheAndaHalf · « **Reply #43 on:** October 23, 2024, 02:40:02 AM »

Quote from: vand on October 22, 2024, 03:44:37 AM

Some day you gotta advance past kindergarten!

Name calling is against the forum rules:
https://forum.mrmoneymustache.com/forum-information-faqs/forum-rules/

MustacheAndaHalf · « **Reply #44 on:** October 23, 2024, 02:56:41 AM »

Quote from: vand on October 22, 2024, 03:44:37 AM

At the risk of feeding the troll...

Quote from: MustacheAndaHalf on October 20, 2024, 11:26:29 PM
Notice the start date: 1972. People pushing gold ALWAYS include 1972-1974, which were 3 of the best 4 years of gold's performance. What gets left out is U.S. leaving the gold standard in 1971, which ended the fixed price of gold. The U.S. won't leave the gold standard twice, so performance from 1972-1974 will not repeat.

1972 - 2024: gold returned 8%/year
1975 - 2024: gold returned 5%/year

(According to Portfolio Visualizer, using portfolio of 100% gold)
https://www.portfoliovisualizer.com/backtest-asset-class-allocation

So what? what's that got to do with the >diversifying< qualities of gold?

As already stated more than once, it's not the >best< outcome that determines the SWR it's the >worst< outcome. A Stock/Gold portfolio was able to sustain a 5.3% withdrawal rate even if you had retired from the very worst time - for both gold by itself, or for the portfolio as a whole. This stuff is worth talking about because it IS more complicated than just looking at the absolute return of individual components from cherry picked years... Some day you gotta advance past kindergarten!

<back on ignore>

Should people buy gold because of your insults? Pointing out the U.S. left the gold standard in 1971 is not "trolling", it is a historical fact. You espouse gold as part of a retirement portfolio, but have yet to acknowledge the U.S. left the gold standard in 1971, nor that the price rises after that were a consequence.
https://en.wikipedia.org/wiki/Gold_standard

Efficient frontier graphs are very sensitive to input data, as they try to weigh the benefits of different asset classes. As I mentioned repeatedly, 1972-1974 should not be used for gold's performance because the U.S. will not leave the gold standard twice (a point vand has never refuted). Here's the problem with those years, when weighing gold vs stocks for diversification:

1973: stocks -18% gold +73% (gap of 91%)
1974: stocks -28% gold +66% (gap of 94%)
1978: stocks +24% gold +127% (gap of 103%)

These are the three largest gaps between gold and stocks from 1972-2024. Notice two of them happened right after the U.S. left the gold standard, and stocks crashed. Including 2 of the 3 largest performance gaps biases the result, making gold look like a good diversifier for stocks. This diversification benefit, a greater than 90% performance gap in each year, can't repeat without the U.S. leaving the gold standard twice.

I provided a historical fact why 1972-1974 will not repeat, and should not be considered for gold in a portfolio. I have shown the diversification benefit won't repeat, and that gold's performance is far lower without the years 1972-1974. All you've learned to do is throw insults, because you're wrong on the facts.

dandarc · « **Reply #45 on:** October 23, 2024, 10:07:17 AM »

Quote from: MustacheAndaHalf on October 23, 2024, 02:40:02 AM

Quote from: vand on October 22, 2024, 03:44:37 AM
Some day you gotta advance past kindergarten!
Name calling is against the forum rules:
https://forum.mrmoneymustache.com/forum-information-faqs/forum-rules/

Counterpoint: that isn't name calling.

dandarc · « **Reply #46 on:** October 23, 2024, 10:08:35 AM »

Quote from: vand on October 22, 2024, 03:44:37 AM

At the risk of feeding the troll...

Quote from: MustacheAndaHalf on October 20, 2024, 11:26:29 PM
<snip because irrelevant>
<back on ignore>

ditto.

SilentC · « **Reply #47 on:** October 23, 2024, 10:57:36 AM »

Another tangent - bonds are not useless. If you blindly buy them they might be, but if you study them they are pretty easy to value. Most people just don’t study them. By easy to value I mean you have lots of history on credit ratings vs rates of default, and spreads (excess yield over treasury) tend to be mean reverting. It was shooting fish in a barrel buying floating rate fixed income in 2023 given where SOFR was anlong with very wide spreads. Now I’ve sold pretty much all my fixed income risk (holding bills) because spreads don’t compensate the risk. I have no idea if the next recession is in a day or a decade, so I’m not market timing, but if I think a bond has a 2% annual chance of default and the spread is 2% or less it doesn’t make sense to own it but if the spread is 5% it makes a ton of sense. Anyway there is a clear framework for incorporating bonds in a portfolio dynamically.

Dicey · « **Reply #48 on:** October 23, 2024, 06:59:45 PM »

Quote from: dandarc on October 23, 2024, 10:07:17 AM

Quote from: MustacheAndaHalf on October 23, 2024, 02:40:02 AM
Quote from: vand on October 22, 2024, 03:44:37 AM
Some day you gotta advance past kindergarten!
Name calling is against the forum rules:
https://forum.mrmoneymustache.com/forum-information-faqs/forum-rules/
Counterpoint: that isn't name calling.

Maybe not, but it is rather rude, which is also covered in the froum rules.

MustacheAndaHalf · « **Reply #49 on:** October 24, 2024, 06:50:58 AM »

Quote from: dandarc on October 23, 2024, 10:07:17 AM

Quote from: MustacheAndaHalf on October 23, 2024, 02:40:02 AM
Quote from: vand on October 22, 2024, 03:44:37 AM
Some day you gotta advance past kindergarten!
Name calling is against the forum rules:
https://forum.mrmoneymustache.com/forum-information-faqs/forum-rules/
Counterpoint: that isn't name calling.

The forum rules state "2. Attack an argument, not a person", which was violated there. The forum rules also show a hierarchy of disagreement, the bottom two being "ad hominem" and "name-calling". It doesn't really matter if the comment is an ad hominem attack or insult - they are both at the bottom.

News:

Author Topic: Pick your diversifiers wisely - rant against bonds (Read 4090 times)