Toward Safe Harbours – Exploring the Bond Portfolio

Posted on October 12, 2019January 16, 2022 by The FI Explorer

Neither a borrower nor a lender be.

Shakespeare, Hamlet

Bonds can provide a safe harbour in times of equity market volatility, and be a critical diversifying element in an investment portfolio.

Yet as the journey to financial independence has progressed, one of the least examined parts of the my FIRE portfolio has been the fixed interest and bond components. This is despite the fact they constituted around a quarter of portfolio assets at the start of the journey.

With the end of ‘big rebalance’ into Australian equities at least in sight, keeping to the target asset allocation may require purchasing bonds or fixed interest instruments over the coming year. This would represent the first significant direct purchase of bond assets since 2014. In turn, this means benign neglect of this part of the portfolio is no longer feasible.

To help establish what this potential future purchase should be, it felt critical to know what I had already in the bond part of the portfolio. The target allocation for bonds is currently set at 15 per cent of the total FIRE portfolio.

This longer read article explores these current bond and fixed interest holdings and seeks to reach a possible choice correct for my personal circumstances and goals. Its focus is not offering advice or fully explaining the operation of bonds (pdf), subjects recently (and better) covered by others.

History of bonds in the portfolio

Bonds have formed at least a small part of the portfolio since its inception. From 2009 to 2014 over $130 000 of new investments flowed into bonds through contributions to a number of Vanguard retail funds. This drove the portfolio allocation to bonds to a maximum of 29.5 per cent of total portfolio assets in July 2014.

There was no particularly deliberate logic or consideration behind this increase, in part it probably reflected inertia from having some regular automatic investments in place, and in part, it also likely appeared a relatively safe haven in the immediate aftermath of the Global Financial Crisis.

This lack of active choices around this asset class is also reflected in the short consideration of bonds in annual reviews, where I simply noted that they were included for diversification and to reduce portfolio volatility.

An overall perspective on changes in the absolute level of bond and fixed interest holdings in the portfolio during the past decade is given by the chart below.

Bond - level stacked - Oct 19 In January 2018 bond and fixed interest holdings reached their highest absolute level of $280 000 or 20 per cent of total portfolio assets. In the portfolio review that year I set a bond allocation of 15 per cent – comprising 5 per cent Australian and 10 per cent international bonds, a division not informed by any particularly strong rationale.

A year later I replaced this with a ‘naive diversification’ approach of an equal split of 7.5 per cent each. As of October 2019, bond and fixed interest holdings are around $257 000 or 15 per cent of the total portfolio.

Examining the stores – what lies beneath the current bond holdings

For the journey so far, my knowledge of what was within the bond holdings of the portfolio was limited. From existing worksheets I knew the split between Australian and global bonds, and the absolute level of each. Beyond that was an undiscovered country.

Part of the issue was finding time to delve into what was a relatively small portion of the overall portfolio. Due to holding bonds through no less than four different Vanguard funds, as well as some other smaller fixed interest holdings through Ratesetter peer-to-peer lending and Raiz, establishing full visibility of the holdings was non-trivial.

Yet a rather complicated multi-step Excel sheet did enable a fuller picture to emerge. Each of the Vanguard retail funds actually invests slightly different proportions in just two underlying wholesale funds:

Vanguard Global Aggregate Bond Fund (which is currency hedged); and
Vanguard Australian Fixed Interest Index Fund

Both of these funds invest in a wide range of bonds and fixed interest investments. These include:

Treasury notes and government bonds
Corporate bonds (including financials, industrials and utilities)
Mortgage-backed securities
Government-related entity bonds
Securitised bonds

These are both extremely diversified. Between them they include bond issuances from over 2 200 different issuing entities, and over 8 000 separate holdings (as issuers normally issue bonds regularly, on different terms or for different durations).

Distributions from the bond holdings through the period have been uneven. An example of this can be seen in the distributions from the Vanguard Diversified Bond fund, which saw no new contributions over the past four years.

Despite this stable balance, in this period distributions for this fund have bounced from as low as $140 per year, to a high of $5270. This is likely to be associated with the funds realising substantial capital growth in periods of lower interest rates, and distributing these gains as part of rebalancing to each funds target allocation.

Looking deeper into the holdings – analysing where and what

Once broken into their component parts, analysis of the actual balance of the bond and fixed interests holdings is possible.

The first thing to note is that while to the overall target allocation of 15 per cent has been reached, the balance of holdings is heavily tilted towards international bonds.

Bond - domint - Oct19 This position is inconsistent with my current naive target of evenly splitting domestic and global holdings. This becomes even clearer when the overall breakdown of total holdings is considered below.

Bond all types - Oct19

This shows that global treasury notes, corporate bonds and mortgage backed securities constitute the majority of bond and fixed interest holdings.

By contrast the two largest Australian holdings – government and government-related bonds – represent only around a quarter of the bond and fixed interest portfolio. Ratesetter peer-to-peer lending represents only 7 per cent of total bond and fixed interest holdings, and without specific action this will continue to diminish as the underlying loans are paid back.

It’s important to note that global bond holdings are themselves further diversified across multiple countries, with the largest holdings being exposed to United States, European Union and Japan.

The full split of all holdings can make it difficult to see clearly the relative weights of different types of holdings. The chart below removes the distinction between global and Australian holdings and simply examines the type of bond or fixed interest holding.

Bondfixed - by type - Oct19 Here some observations can be made.

Government or government-related entity debt is the single largest component. Corporate bonds make up just under 20 per cent of holdings, with additional small holdings in mortgage-backed securities and peer-to-peer lending.

The question then looms, is this the right mixture of holdings, and how would an answer to this be established in current market conditions?

I have never been able to get this business of loans and interest into my head. I have never been able to understand it.

Philip II of Spain to his Minister of Finance, 1580

Becalmed – buying and holding bonds in a ‘zero rate’ world

If reaching the portfolio target asset allocation requires a purchase of bonds in the near term, in many ways this will be a difficult move to rationalise and execute.

There is a plausible case that Australia and the world are at the end of a four decade secular bull market – or even bubble – in bonds. Since the early 1980s across most developed markets bond yields have fallen to historically low levels.

Significant portions of government bonds around the world, and even some corporate bonds, are currently trading at negative yields. That is, borrowers are paying for the privilege of loaning their money out to debtors. One day, this extreme and unprecedented global trend could reverse. Bond yields could revert to levels closer to, or even above, their historical average.

There is, however, a significant chance that bond yields seen in the past will not be seen for decades to come. There is also the chance that further falls in yields could occur, leading to further gains in the capital value of bonds. Which of any of these potential futures will play out is difficult to forecast.

Yet this is not the only consideration. The original function of bonds in the portfolio is to reduce volatility, and therefore the primary consideration is not simply their absolute performance, but rather how their returns can be expected to move in relation to other parts of the portfolio.

Traditionally bonds have had low correlations to equity returns, which is a critical consideration given the set 75 per cent target allocation to equities in the portfolio. This has not been uniform, so for example I personally recall 1994 being an exceptional year in which both bonds and equities experienced sharply negative returns. An example of this record can be found in this calendar year return comparison of Australian bonds and equities here (see p.19).

It is possible that this correlation could break down into the future, although there are some underlying drivers for bonds being less volatile than equity, it is perfectly possible for these to be offset by external conditions for prolonged periods.

All bonds are not created equal

It is also worth noting that different types of bonds will have different correlations with risky equities. For example, a government with taxation powers and the option of printing local currency can technically never be forced to default on payments of domestic government bonds, regardless of equity market conditions.

By contrast, although corporate bond holders stand in front of equity holders in the queue for available company cash in difficult times, some of the same market forces could easily be expected to exert themselves on a corporations capacity to pay debt and provide equity returns, giving them higher correlations.

Similar considerations apply to the risk characteristics of peer-to-peer lending – which can often be used for car loans, small renovations or other similar purposes. Risks to repayments of these loans are likely to be significantly correlated to equity market conditions, and to changes in employment and wages conditions. Thus they may produce lower diversification benefits (even where the income is attractive) than traditional bonds.

Flags of convenience – are there benefits in holding offshore bonds?

A major finding from reviewing the detailed holding is the level and breadth of foreign bond exposure. Yet aside from the not insubstantial psychological satisfaction of apparently holding the equivalent around $200 in Romanian or Panamanian bonds, is there actually any financial benefit in holding foreign over domestic bonds?

This question is the focus of an excellent 2018 Vanguard research paper (pdf) Going global with bonds. This analysis uses 30 years of historical data to make observations on the value of global diversification in bonds. It finds that:

Global exposure reduces volatility. Global bonds hedged into the local currency had significantly lower volatility than holding just domestic bonds, across major developed nations including Australia.
In part by reducing individual bond risks. Global bond holdings reduced exposure to local market risk factors and gave better diversification to a range of interest rate and inflation risks that impact bond returns.
Hedging global bonds smooths volatility. Whilst hedged and unhedged global bond returns will be similar over long periods, hedging historically smooths volatility – indeed, significant unexpected falls in the Australian dollar would be needed to mathematically justify not hedging.
And offers more downside protection. Hedged bond portfolios have offered better protection in adverse market conditions than their unhedged equivalents.
Which can help reduce risk even in broad portfolios. Adding hedged global bonds to a mixed equity and bond portfolio can offer some modest portfolio volatility reduction (of around one per cent).

A necessary caveat is that these findings are a function of the time period selected. Yet the analysis is also the best empirical review I have seen so far on the allocation issue which takes into account recent and prevailing bond market conditions.

Departing course – which types of bonds should the portfolio include?

Finally, there is the question of whether the portfolio should seek an alternative exposure to different types of bonds and fixed interest instruments than that currently held. In principle, under modern portfolio theory, there is a specific bond portfolio that would minimise risk and produce an optimal risk-adjusted return for every different equity portfolio and investment target.

Yet having adopted a market capitalisation weighting approach for equities, through equity index ETFs such as Betashares A200 and Vanguard’s Australian equities ETF (VAS), it would not be consistent to start seeking to make active sectoral ‘bets’ in types of bonds and fixed interest. Especially if there was no reason to believe a special insight or information would enable this active approach to reliably add value.

At present, therefore, being ‘market capitalisation weight neutral’ (i.e. holding different bonds in roughly the proportion of their total market value) is a more efficient approach given the low likelihood of outperforming professional bond market participants and higher risks.

At the time of writing, with a gradual reduction in Ratesetter balances expected to continue, the bond portfolio is allocated quite close to global capitalisation weighted benchmarks (i.e. it reflects the size and make up of bond and fixed interest markets globally). This is an automatic result of how Vanguard’s funds themselves generally seek to match Australian and global sector weightings.

Summary – applying the learnings

The exploration of the bond and fixed interest part of the FIRE portfolio has filled in a lot of knowledge gaps that should not have existed. Yet what I have found provides some confidence that thankfully this past neglect has not come at a significant cost or increased risk.

From the review a few useful points and lessons have been reinforced. To summarise:

Reliance on Vanguard retail funds has built the foundations of a diversified portfolio. The bond portfolio is already well-diversified across different bond and fixed interest issuers, markets, countries and risk types.
A ‘home bias’ for Australian bonds and fixed interest is not warranted. Based on the data and analysis above, there is no reason to target a level of Australian bond holdings at anything above around 1.5-2.0 per cent as there are no significant advantages of any ‘home bias’.
Future bond investments will be made through market capitalisation weighted index vehicles. This could plausibly include Vanguard’s Hedged Global Aggregate Bond Index Fund (VGBND) or further investments in Vanguard’s Diversified Bond retail fund.
Peer-to-peer lending should be considered as a separate type of fixed interest asset to traditional bonds and fixed interest. While peer-to-peer lending can have diversification and income benefits, it may not be a direct bond or fixed interest substitute. That is, it may not provide significant reduction in portfolio volatility should loan defaults rise in a downturn.

With these points in mind future decisions on investments in bonds will be made with increased knowledge and taking into account recent market evidence. And, importantly for the decision ahead, the contents of the stores below deck are known and accounted for.

Sources and further reading

Bernstein, W. The Intelligent Asset Allocator, McGraw-Hill, New York, 2000

Graeber, D. Debt: The First 5000 Years, Melville House, 2011

Macdonald, J A Free Nation Deep in Debt, Farrar Strauss and Giroux, New York, 2003

Vanguard Plain Talk Library Bond Investing (pdf)

Vanguard Research Going global with bonds: The benefits of a more global fixed income allocation, April 2018 (pdf)

Disclaimer

This article does not provide advice and is not a recommendation to invest in any specific bond or fixed interest instrument. Its sole purpose is to discuss bond and fixed interest investment issues relevant to my personal circumstances.

Setting of the Sails – Role of Gold and Bitcoin in the Portfolio

Posted on July 20, 2019January 15, 2022 by The FI Explorer

One ship drives east and another drives west
With the self-same winds that blow.
‘Tis the set of the sails
And not the gales
Which tells us the way to go.

Ella Wheeler Wilcox, The Winds of Fate

Future returns are unknowable with any degree of precision. A portfolio must contend with all that future market prices and developments put before it, whilst seeking to earn the best possible return for the level of risk assumed.

This uncertainty is a core issue for portfolio design. Part of my approach to building my FIRE portfolio has been to target a small allocation to alternatives such as gold and Bitcoin to deliver reduced portfolio volatility, and improved returns. My current target allocation set earlier this year is 7.5 per cent gold and 2.5 per cent Bitcoin. This post explores the reasons for, and basis of, this approach.

Portfolio design – one wind, different directions

In designing the FIRE portfolio, the key guiding principle has been maximising the overall risk-adjusted return, whilst minimising unnecessary volatility.

The important implication of this is that it is not the performance of the individual portfolio parts that I am trying to maximise. Rather, it is the performance of all of the component parts as they interact that is of prime concern.

The objective is for the mix of all of these different holdings to play their part together to enhance portfolio returns or reduce volatility. Decisions on asset allocation – or the mix of assets held – has been repeatedly been shown in academic studies to explain around 90 per cent of the volatility of portfolio returns.

This approach is consistent with the simple guidance to diversify. Underlying it, however, are some observations of modern portfolio theory and the Capital Asset Pricing Model, that can be summarised in the following insights:

the investor should seek to mix assets with non-correlated returns (i.e. returns that move in different directions) to achieve an optimum balance of likely returns and portfolio volatility
not all extra risk taken by an investor is automatically compensated by higher returns
the investor should consider each additional investment security or asset from the perspective of how it will contribute to overall portfolio risk and return

At any given time this can mean that one ‘wind’ will send the individual portfolio components in different directions. In short, the approach is not one that will deliver a portfolio without any losses or low returns in the set of assets held at any given time.

Asset correlation – assessing the crosswinds

The critical ingredients for the approach to be effective are assets that do not move together – that is, uncorrelated assets. A traditional example used in portfolio design are equities and bonds, which have over time often tended to move in opposite directions (e.g. be inversely correlated) in many markets. This is the basis for traditional investment guidance to include greater bond holdings to dampen the volatility of equities.

Gold has tended to have a low correlation to other asset classes. An example of the effects of this on equity portfolios is described in this research paper (pdf) – from the World Gold Council – which found that adding gold holdings to an all equity portfolio both lowered the volatility of returns and increased total returns over the 1968-1996 period (see p.47 and Figure 4.6). The academic evidence for the low correlation of gold to equity returns is, in fact, strong over multiple periods.

Moreover, this diversification benefit appears when most needed. As this recent paper in the International Review of Financial Analysis notes:

…we think that a review of the results from earlier papers on this issue,
coupled with our findings, points to the fact that gold is always a hedge or, at
worst, always an excellent diversifier of portfolio risk. Gold’s usefulness in
managing risk does not disappear in a crisis when the prices of the vast
majority of assets tend to be perfectly correlated. (He, 2018)

That is, gold seems to generally hold up as providing non-correlated returns, even when extreme market conditions prevail. Globally, central banks – including Australia’s Reserve Bank – also seem to recognise this characteristic. It is in part why central banks collectively own around 17 per cent of gold currently above ground.

Setting the level of gold exposure – competing evidence

There is considerable discussion and debate on the right level of gold holdings to maximise the diversification benefit, and few definitive answers.

The optimum level will vary under most estimation approaches, which inevitably are based on models that build on historical observed relationships and correlations. These correlations themselves vary over time and between markets and countries.

An original study by Jaffe for institutional portfolio managers recommended a 10 per cent allocation against a basket of international equities. Additional studies (pdf) by other authors have recommended 9.5 per cent, and between 0.1 per cent to 12 per cent depending on which country the investor is in. As an example, the country-specific weights typically fell within 3 to 8 per cent for developed countries.

More complex methods than classical mean variance analysis, which take into account the positive skew of gold returns, produce different results again. A 2006 study which examined 1988-2003 data recommended a holding of 4-6 per cent under classical portfolio optimisation approaches, but a lower figure of 2-4 per cent taking return ‘skewness’ into account.

Diversification and Bitcoin – looking at the record

My purchase of Bitcoin began as an exploration of a new financial technology driven by curiosity. The present question is, however, does it deliver any additional diversification benefits beyond gold holdings?

Conceptually, Bitcoin can be said to share some characteristics with gold that might be expected to reduce any diversification benefit. They both represent highly liquid assets that when held personally are no other parties liability. They are not issued by central banks or other monetary authorities, and they can be transferred. So is there a case for holding just one or the other?

The tentative answer is that despite some conceptual similarities, they do appear to behave differently.

So far, in the decade between July 2009 and February 2019, Bitcoin has shown a low positive correlation to gold (see In Gold We Trust (pdf), p.245). This is consistent with my own observations in my portfolio in the last three and a half year period, with a low correlation of 0.1 over the entire period in the chart below. Bitgold correl

On its face it appears Bitcoin may well be a useful complementary alternative holding, offering diversification benefits distinct from other combinations of holdings.

Unlike gold, there is not a clear empirical or academic basis for setting a ‘right’ level of exposure to Bitcoin. The recent In Gold We Trust report (pdf) discusses and analyses one possible approach – a 70/30 split between gold and Bitcoin, indicating that this delivered similar maximum drawdowns to a gold only portfolio, but with higher returns. Yet this finding is only a function of the extraordinary positive returns from Bitcoin to date, and may not be repeated.

Trade-offs, risks and limits of exposure to alternatives

There are acknowledged trade-offs and risks to investing in alternatives such as gold and Bitcoin.

First, they produce no income or cashflow. Their return is based entirely on capital gains. This is often cited as a definitive proof that they do not represent part of any proper investment portfolio.

Yet, as a part of a portfolio, alternatives can reduce the absolute volatility of the capital value of the portfolio, and – historically in the case of gold, can also increase overall returns. Given final capital value and returns over time are critical inputs into FI, these characteristics are relevant and worth considering.

A potentially stronger objection is that while alternatives may have been useful in the past, they cannot be guaranteed to be so in the future.

That is, the correlations and diversification benefit that has been observed, may disappear. This is entirely possible, and ultimately unknowable. The diversification benefits of gold have a far longer history. Its roles in industry, manufacturing and jewellery would seem likely to continue to guarantee that at any given time there will be some minimum demand for gold, and a relationship between its price and other asset prices that is not perfectly correlated.

For Bitcoin, the same cannot be said. There are many plausible scenarios in which Bitcoin’s value declines, it falls in usage, and becomes the equivalent of niche digital collectible with little residual value.

The disappearance or long-term reversal of ‘known truths’ in finance is not impossible. There are significant periods in capital markets in which bonds outperformed equities, negative yielding debt has moved from something previously unobserved, to a commonplace across many world bond markets. By some measures, global interest rates are at 5 000 year lows. Few developments should be dismissed as inconceivable looking forward.

This suggests that any analysis based on historical trends should be relied on with modest expectations around its accuracy. Yet importantly, this applies not just to speculation around the role and benefits of alternatives. It also applies to traditional investment classes, such as equities or bonds.

For example, the continuation of a positive equity premium for Australia, or any other nation, is not foreordained. Australia’s comparatively high equity returns are in fact an anomaly looking across developed countries. There are no particularly strong reasons to suggest this will necessarily continue.

Set of the sails – applying the evidence to a FIRE portfolio

The role of gold and Bitcoin are primarily as non-correlated financial instruments for diversification, and as an insurance against extreme capital market events. No actual positive return is assumed for either asset. The evidence discussed above leads me to the following conclusions, for my personal circumstances and risk tolerance.

Reliance on equities as the engine for portfolio growth. Long term equities continue to have a strong record of providing higher total returns, earning their place as the centrepiece of the portfolio.
Reliance on history of performance of gold to reduce volatility. Some exposure to gold appears to reduce volatility and potentially enhance returns historically, making it a potentially beneficial addition to my FIRE portfolio.
A small role for gold based on tested academic evidence. Past evidence suggests a gold allocation of between 5 to 10 per cent is sufficient to capture diversification benefits, without compromising long-term portfolio returns
With Bitcoin potentially adding further diversification. Bitcoin appears to be non-correlated to equities, bonds, and gold, meaning it potentially is a useful further additional source of diversification benefit.
But with modesty about what the future holds. Aside from Bitcoin being volatile, there is an inadequate history to know how it will perform compared to other assets through a full cycle, or whether it has a long-term future.
Recognising the limits of knowledge and history. Asset performance, diversification benefits, volatility and returns which are historically based can and do reverse at times, meaning the ‘best’ portfolio will only ever be known in retrospect.

The alternatives target allocation set earlier this year is 7.5 per cent gold and 2.5 per cent Bitcoin. As of July 2019, a strict reading of these targets suggests I need to moderately lift my exposure to gold, and sell approximately 75 per cent of my Bitcoin holding.

I currently plan to do neither of these things. This is because:

The volatility of Bitcoin is such that ‘chasing’ a target allocation by buying and selling is likely to incur high transaction costs (including realising capital gain tax).
A plausible scenario is the apparent over-allocation to Bitcoin resolving itself through substantial price declines as previously experienced (at its previous low, the allocation was close to the 2.5 per cent target).
Similarly in the case of gold, both price volatility and the goal of minimising transaction costs suggest it is better to seek to adjust holdings only when they fall well outside the target allocation for a sustained period.
The overall size of the entire alternatives allocation (a 10 per cent target) is more significant than the individual sub-targets.
Before making new investments to pursue my portfolio allocation I perform a ‘with and without’ test, notionally removing the Bitcoin holdings for a moment from the portfolio, to identify if recent fluctuations in the value of Bitcoin are driving a perverse allocation choice which would be entirely different were it not for Bitcoin. While not theoretically ‘pure’, this is a pragmatic adaptive approach that recognises the lack of clear history and knowledge about the portfolio behaviour and characteristics of Bitcoin.

So the sails are set, and the wind will come. These settings allow me to feel that whatever direction they happen to blow, there is the best chance possible based on evidence that they will help in the journey that remains.

Sources

In Gold We Trust 2019 – Extended Report

Harmston, S. Gold as a Store of Value, Research Study No.22, World Gold Council, 1998

He, Zhen et al. “Is Gold Sometimes a Safe Haven or Always a Hedge for Equity Investors A Markov-Switching CAPM Approach for US and UK Stock Indices”, International Review of Financial Analysis, Vol. 60, October 2018

O’Connor, F et al. “The Financial Economics of Gold – A Survey” in International Review of Financial Analysis 41 · July 2015

Disclaimer: This article does not provide advice and is not a recommendation to invest in either gold, Bitcoin or any alternative assets. Its sole purpose is to provide an explanation of why – in my personal circumstances – I have chosen this exposure.

On Measurement – A History of Financial Benchmarks

Posted on June 15, 2019February 15, 2021 by The FI Explorer

If you have built castles in the air, your work need not be lost; that is where they should be. Now put the foundations under them.

Henry David Thoreau Walden

Essential to navigating any course is knowing your intended destination, and measuring one’s current position. Since the commencement of my journey to financial independence there has been a rather restless search for, and refinement of, the right measures to use.

Knowing just how far one is through a journey also helps to provide a sense of momentum. Key to measuring how much progress has been made is consciously thinking about and defining the end point.

This piece will look through the various benchmarks and measurements adopted, retained and discarded through the my journey, to share what I have learnt about measuring progress to financial independence, and set out the logic behind my current portfolio goals.

Early benchmarks – There lies the port (1999-2006)

There is a strong argument that the first FI benchmark I ever set was the most correct, and the rest have been excursions.

In September 2004, after around three years of consciously building up an investment portfolio, I read perhaps the most foundational financial independence work, Your Money or Your Life.

A central recommendation of this book is to physically graph out monthly income and expenses, from which naturally emerges a measure of the proportion of passive investment income to actual expenses. I did this from June 2003 until December 2005. Each week my calculations produced a simple percentage result of how close I was to the ‘crossover point’ – that is, the point at which passive income exceeds monthly expenditure and financial independence is achieved.

By the time I abandoned the monthly record I was, by my rough calculations, around 80 per cent of the way to the ‘cross over’ point. This was soon to fall, however, with the significant purchase of a house to live in with part of the portfolio, and a continuing fall in interest rates affecting some cash-based savings.

This same overall approach, however, underpinned my first ever explicit benchmark set for the portfolio in 2006. It was a quite logical and simple one: yearly investment returns should match my own average yearly expenditure.

Measures multiplied and buried (2007-2016)

As a measure this was hard to improve on, but this provided no deterrence. With the investment portfolio starting to grow after the house purchase I thought about what would be meaningful benchmarks again, and came up with a layered approach.

There were three new benchmarks under this approach, broadly themed as:

Comfort – This benchmark was investment income to achieve the median average national salary, with progress estimated as a percentage of current portfolio value against the total portfolio needed to deliver this at an assumed rate of return.
Independence – This measure was to achieve investment income equal to the median salary of a Federal public servant serving at a mid-range executive level. This was chosen because it represents a very comfortable standard of living and an average level of career achievement for a person with my qualifications.
Freedom – This measure was for investment income to be equal to the salary of a median public servant at a more senior level in a central government agency. This would enable a lifestyle that is untroubled by material need.

The logic behind this triple standard was a recognition that just reaching a comfortable level of investment income did not necessarily encompass all of my aims. Rather, I was curious to understand what might lay beyond, and what resources might give me a lifestyle indistinguishable from most of my peers in terms of ability to fully engage with the world’s opportunities, without requiring paid work.

The decision to measure progress by reference to an external benchmark was due to two factors. First, a benchmark that is externally linked to a particular standard of living helps ensure that the goal shifts broadly with changes in the broader community. Second, it made accounting for inflation easier, as the benchmark already accounted for its effects.

Years after these three benchmarks were set, I read about the six levels of financial independence, and it was apparent that as well as drawing on some similar concepts in Your Money or Your Life I had unconsciously replicated some of these. The six levels represent an excellent framework through which to think about the different stages of financial independence. Others, such as the ChooseFI podcast, have usefully added other milestones (i.e. ‘half-FI’) to supplement the approach, and place some way markers in between some of the larger steps.

The precise numerical expression of these three layered objectives shifted through time as I learnt more about realistic return expectations and updated them for the impacts inflation. In July 2007 I set a target portfolio value target of $750 000, with the explicit – though ultimately unrealistic – expectation of it producing around $50 000 in annual portfolio income. The goal of providing for a stream of passive income specifically targeting an average income (of $58 000) can be traced back a decade to July 2009.

By 2010 I had estimated that a portfolio of around $1.1 million would be required to meet this average income benchmark. I updated this target to reflect more realistic information and evidence on likely sustainable returns in 2016, first setting my previous FI target of $1.47 million.

Finding a benchmark, or a measure of the journey, was therefore an iterative exercise – first begun and then improved on as I learnt more. Along the journey I have tracked, and in some cases continue to track, a number of other metrics. Some of these numbers fall out of existing spreadsheets, others are historical relics in little used Excel workbooks – seeming important for a time, but now neglected and overtaken as meaningful marks of progress. These other metrics include:

Asset reserves in weeks. A measure essentially of how long I could last if all employment income stopped tomorrow. This was a significant early metric, and was a comfort to review from time to time. To be able to note that if the worst came to the worst, it might be 6 or 12 months before I could not meet expenses gave a positive feeling of a basic level of security.
Passive income expressed as numbers of hours worked at minimum wage. How the portfolio income compared to the Australian minimum wage, i.e. how many hours ‘free work’ did the portfolio complete on my behalf? This is a way of thinking about the additional income a portfolio has produced at no physical cost, to consider the hours of work your dollars are putting in which you do not have to, boosting your financial progress.
Remaining deficit to FI target. This is simply the ‘distance still to travel’ number, and towards the mid and late stages of the journey can be more motivating and tangible to focus on than the long progress already made. At this stage, forward progress week to week might be almost invisible in percentage terms, and yet the absolute deficit can still be closing proportionally faster.

Current navigation aids (2017-2019)

My current approach is to keep benchmarking against external standards, but to supplement these with some specific personal FI benchmarks.

In January 2019 I reset my two external benchmarks of progress (Objectives #1 and #2).

Objective #1 is a passive income benchmark that is equal to the the median annual earnings of an Australian full time worker ($67 000). That is, approximately 50 per cent of workers earn both less and more than this figure. This is drawn from Australian Bureau of Statistics earnings data, which is updated at least annually, and which therefore can be consistently tracked through time. My logic for picking this benchmark is that any reasonable concept of ‘enough’ should encompass and be somewhat anchored around the earnings of an average worker. To have access to this income, without a single hours paid work being required, represents a significant achievement in freeing oneself all of the potential cares of working career.
Objective #2 is set at the approximate equivalent of average Australian full-time ordinary earnings ($83 000). As an average, this ABS benchmark is skewed upwards by a small number of higher earners. This second longer-term goal is designed to reflect a more ‘business as usual’ lifestyle reflecting my personal circumstances. At least in my current phase of life, the lower income of Objective #1 would effectively represent rather than more of a ‘leanFIRE’ concept. As I have previously observed, the income assumed in Objective #2 is closer to the level of expenditure at which I think I would truly become indifferent to working or not.

I have also started tracking these any other measures both against the FI portfolio, but also against an expanded ‘All Assets’ portfolio. This recognises that I have some significant superannuation assets that currently sit outside of the investment portfolio.

This means I now seek to assess progress on two different bases: first, the current measure based on reliance on the investment portfolio alone and second an ‘All Assets’ measure with superannuation assets taken into account.

The reason for this dual approach was that it was artificial and distortionary to my own thinking about the issue to entirely ignore a substantial potential contributor to a FI target in the form of superannuation, even if it comes with accessibility restrictions and some legislative risk.

Due to these risk and restriction factors, I plan to continue to target financial independence through my private investment portfolio alone, with superannuation providing an additional margin of safety and buffer.

IMG_20170930_130516_249

Cross-bearings and lines of position

My other recent change was to report against an expanded set of benchmarks, beyond just my formal investment objectives.

Since January 2019 I have reported against two additional measures. First, my average annual credit card expenditure (a ‘credit card FI’ benchmark), and the second is an aggregated estimate of total current annual expenditure.

The credit card purchases measure is a way of keeping my financial progress grounded in the reality of what I actually spend. It is currently set at $73 000 per annum, equivalent to the past 12 months of credit card bills. The measure is derived by calculating how much of this expenditure the portfolio is – using the assumed real return rate of 4.19 per cent – producing in income.

This has the benefit of both automatically tracking broad spending trends and adjusting for the inflation I personally experience through time. It is also a highly salient measure. As one stands in front of a paywave machine, it is some comfort to think that portfolio income is paying over 75 per cent of the bill.

As an additional measure I also track actual month to month credit card purchases, and compare them again either current distributions or a 3 year rolling average (as illustrated below).

Measuring - 3yr card The total income measure is quite approximate and results from adding some known fixed expenses (such as rates and utilities) that I do not pay through credit card to my total credit card expenditure. It currently totals $96 000. As I have noted, I recognise that it is by no measure a frugal existence, and my good fortune in being able to live in this way.

An example of these measures is given below, using the portfolio position on in the recent May Monthly Portfolio Update as inputs in this case.

Measure	Portfolio	All Assets
Objective #1 – $1 598 000 (or $67 000 pa)	100.0%	137.3%
Objective #2 – $1 980 000 (or $83 000 pa)	80.7%	110.8%
Credit card purchases – $73 000 pa	91.8%	126.0%
Total expenses – $96 000pa	69.8%	95.8%

Future measures – the end of reckoning?

So what then for the future of benchmarks in measuring the portfolio?

For the moment, the present measures seem sufficient. Recently, however, I have added some additional metrics to watch as the portfolio changes in value.

These new ‘watch’ metrics are the required safe withdrawal rates implied by drawing each required income (i.e. $67 000 or $83 000 as per Objective #1 and 2) from the portfolio. That is, taking the target income levels as fixed, and then calculating what percentage of the portfolio this represents.

Mathematically, this is just a re-arrangement of the method of determining the level of income from the portfolio, but not assuming the current rate of return of 4.19 per cent. So it is equal to the required benchmark income divided by the Portfolio total (so for example, Objective #1 income $67 000/Example Portfolio Level of $1 430 000= 4.68 per cent).

This metric helps make visible exactly the level of investment returns (or safe withdrawal rate) that would be implied by a total reliance on the portfolio at this moment. The reason this is helpful is that a significant set of academic and other analyses cover the issue of the inverse correlation of safe withdrawal rates and equity market cycles.

Put simply, a higher safe withdrawal rate is riskier at time of expensive market valuation (pdf), i.e. good times, than after equity market falls. Conversely, low safe withdrawal rates may be marginally ‘safer’ following substantial equity market falls.

Safe withdrawal rates are typically designed to not fail given a long backtested history of actual market movements over a range of conditions. Yet there is still value in eyeballing the assumed safe withdrawal rate as a cross-check on any decision to cease paid work, and feeling comfortable with that figure.

Observations – Finding a True North

The power of setting goals and benchmarks cannot be underestimated. My own observations on the process of measuring progress to FI goals are summarised below:

Starting is better than finding the best measure. Overall though I have found and discarded many measures and goals along the way, but the choice to start to measure and hold myself accountable for progress was a powerful motivation and tool. Each measure and benchmark helped in its time.
Different measures may serve you at different times. Linked to the above, different measures will seem relevant and motivating through different stages of the journey, whether it be ‘reaching zero’, a saving rate or progress towards a specific FI number. Those changes in which measures seem the best fit may actually be important markers of the changing phases of the journey
Inflation should be accounted for in any measure. With inflation at historic lows, this may seem unimportant, but with apologies to Trotsky: ‘You may not be interested in inflation, but inflation is interested in you’. FI measures that don’t account for the impact of inflation on purchasing power over years and decades ahead are dangerous to your future lifestyle and goals. Whether it is updating nominal dollar targets regularly for inflation, or exclusively using ‘real’ dollars and rates of return alone, it is critical that goals account for inflation impacts.
Measures will be personal choices. The right measures will be deeply personal, influenced by circumstances, preferences, and future goals. There is not likely to be one ‘right’ set designed just for you, even though many of the most common measures (the 4 per cent safe withdrawal rate ‘rule of thumb’ or savings rates) have sound logic behind them.
Choose measures that make you consider the whole picture. It is possible to fixate on a single measure for clarity, and for this to provide only a narrow or incomplete view of progress. So behavioural ‘framing’ impacts should be considered when setting measures. That is, consider what the measure might obscure or hide, and its impact on your choices. Examples might be: assets left out of consideration in net worth style measures. Ideally, between them the measures adopted should provide a holistic picture of overall progress, without distorting decision-making by leaving out important aspects of your financial decision-making or circumstances.

Set and Drift – Estimating Future Income from the Portfolio

Posted on May 18, 2019January 15, 2022 by The FI Explorer

Cultivate an asset which the passing of time itself improves.

Seneca, Letters XV

The focus of the voyage to financial independence so far has been designing the portfolio, and measuring the distance still to travel. There is more basic question to be asked as the journey progresses – will the portfolio produce the income targets set for it, or will something need to change?

Currently, the income estimates from the portfolio targets – $67 000 from a short-term target of around $1.6 million and $83 000 from a target of around $2.0 million in several years – are set on an assumption of a total portfolio return of 4.19 per cent.

That does not mean, however, that the portfolio will simply automatically produce an income of that level. Just pointing the ship in the direction of travel is not enough. This is because the total return assumes both capital growth and distributions or interest.

This analysis examines what income the portfolio is likely to produce when the targets are achieved, and assesses whether or not selling down or changing the portfolio in other ways to meet the income goals may be necessary.

To answer this question, history and three different methods of estimating the potential income produced by the portfolio are reviewed.

Approach #1 – Navigation by landmarks

The first approach is to simply use what is already known to establish one’s position.

Previous analyses have discussed the overall trends in portfolio distributions, and reached some approximate estimates of the likely underlying level of distributions. These estimates differ according to the precise method chosen, and time period considered. So far, these analyses have established that the portfolio appears to be generating between:

$5 000 per month or $60 000 per year, if an approach where the moving average of the past three years of distributions is used; or
$3 800 per month or $45 000 per year, if a conservative approach of an average of the past four years of distributions is applied.

This is healthy progress, however, both of these figures are short of the Objective #1 income requirement of $67 000 per year (or $5600 per month), and even further from the projection of $83 000 (or $6 900 per month) under Objective #2.

Will the future look like the past?

These historical figures are useful because they are real data based on holdings in the actual portfolio. Their disadvantages are that they are backward-looking. This has two possible impacts.

First, the growth of more than 50 per cent in the total portfolio size over even the past three years means that the level of historical distributions will underestimate the income generation potential from the now larger portfolio. In short, this is like trying to estimate interest from a bank account by looking at your balance three years ago.
Second, the distributions of three or four years ago will reflect past asset allocations, and investment products. As an example of this, two years ago the portfolio contained over $55 000 invested in Ratesetter’s peer to peer lending platform. This was earning an average income return of 9.1 per cent. Today, Ratesetter is less than half of this size, due to a slow asset reallocation process and withdrawals as loans mature.

This suggests a purely backward view of the actual achieved distributions may be incomplete and misleading.

Taking an average distribution rate approach

The other potential way of estimating the income return of the portfolio is to use the average distribution rate of the portfolio in the past.

The rate is calculated as total distributions over a defined period divided by the average portfolio level over the same period.

This eliminates any errors from the first impact discussed above of growing portfolio growth size, as it is a rate rather than a level measure. It does not eliminate the second impact. For example, higher interest rates meant that cash holdings in 2013-14 could make up over third of total distributions, a position not likely to reoccur in the short or even medium term.

Yet it still may be an approximate guide because while overall portfolio asset allocation has shifted in the past two and half years, it has remained within some broad bounds. As an example, total equity holdings were at 70 per cent of the portfolio both 5 and 10 years ago. Additionally, using a median long-term average of 4.4 per cent will tend to reduce the impact of one-off changes and outlier data points.

As established in Wind in the Sails the average distribution rate over the past two decades has been around 4.4 per cent.

SAD Dist Average This implies that the portfolio would produce:

$5 900 per month or $70 300 per annum income when the portfolio is at Objective #1 (e.g. this suggests that the target income at Objective #1 would be met, with around $3 000 to ‘spare’).
$7 300 per month or $87 100 per annum income when the portfolio is at Objective #2 (e.g. as above it suggests meeting Objective #2 would produce around $4 000 more income than actually targeted).

An interesting implication of this is that the portfolio has been producing distributions (at 4.4 per cent) at a rate that is higher than the overall rate of assumed long-term total return (around 4.2 per cent).

This is consistent with the fact that the Vanguard funds, and to some extent shares and other ETFs have been realising and distributing capital growth, not just income. This means that if I truly believe my long-term total return forecast is more accurate than the distributions estimate, I would need to re-invest the difference, to ensure I was not drawing down the portfolio at a higher rate than intended.

Approach #2 – Navigation by ‘dead reckoning’

A different approach to reaching an income estimate from the portfolio is to forget about the actual history of the portfolio, and look to what the record shows about the average distribution rate from the asset classes themselves.

That is, to construct an hypothetical estimate of what the portfolio should produce, based on external historical data on average income from the individual portfolio components of Australian shares, international shares, and fixed interest.

To do this, estimates of the long-term income generated by each of the asset classes in the portfolio are needed. For this ‘dead reckoning approach’ I have used the following estimates.

Table 1 – Asset class and portfolio income assumptions

Asset class	Allocation	Estimated income	Source
Australian shares	45%	4.0%	RBA, 1995-2019, May Chart Pack
International shares	30%	2.0%	RBA, 1995-2019, May Chart Pack
Bonds	15%	1.0%	Dimson, Marsh and Staunton Triumph of the Optimists 101 Years of Global Investment Returns, Table 6.1
Gold/Bitcoin	10%	0%	N/A
Total portfolio	100%	2.55%

This analysis suggests that at the target allocation for the portfolio, based on long-term historical data, it should produce an income return of around 2.6%.

This equates to:

$3 400 per month or $40 700 per year when the portfolio is at Objective #1
$4 200 per month or $50 500 per year when the portfolio is at Objective #2

These figures are also well short of the income needs set, and so imply a need to sell down assets significantly to capture some of the portfolio’s capital growth.

Abstractions and obstructions

Of course these figures are highly averaged and make some simplifications. Year to year management will not benefit from such stylised and smooth average returns. Income will be subject to large variations in distribution levels and capital growth will vary across asset classes and individual holdings.

Another simplification is that is analysis does not include the value of franking credits. If it is assumed that Australian equities continued to pay out their historical level of dividends, and the franking credit rate remains at the historical average of around 70 per cent then Australian shares dividends should yield closer to 5.2 per cent, lifting the total income return of the portfolio to around 3.1 per cent. In turn, this would marginally reduce the capital sell-down required. Adjusting for this impact means the portfolio income would be $4100 per month at Objective #1, and $5100 per month Objective #2

Yet these assumptions can be challenged. It is possible that the overall dividend yield of the Australian market will fall and converge with other markets. This would be particularly likely to happen if further changes to dividend imputations or the treatment franking credits to occur. It could also occur due to a maturing and deepening of Australian equity markets and domestic investment opportunities available to Australian firms. Shorter term, uncertainty around the future ability of shareholders to fully benefit from franking credits could encourage a payout of credits currently held by Australian firms.

Approach #3 – Cross-checking the coordinates

Due to these simplifications and assumptions, it is appropriate to cross-check the results of one method with other available data. An alternative to either a purely historical approach using distributions received, or the stylised hypothetical above discussed in Approach #2, is relying on tax data.

Specifically, taxable investment income can be estimated as the sum of the return items for partnerships and trusts, foreign source income and franking credits (i.e. items 13, 20 and 24) in a tax return.This has been previously discussed here.

Using this data is – of course – not independent of my own records of distributions. Its benefit is that it strictly relies on verified data provided in tax calculations. This will include income distributions and realised capital gains from within Vanguard funds, for example, but will not pick up unrealised capital gains.

As with Approach #1, as the portfolio has changed in size and composition the absolute historical levels of taxable will not necessarily produce the best estimate of the expected level of distributions looking forward. For example, because it is drawing on a period in which the portfolio was smaller, a five year average of investment income would suggest future annual investment income of $32 300 or $2 700 per month.

So instead an ‘average rate’ approach can be used to overcome this. Over of the past five years, the portfolio has produced an annual taxable investment income of around 3.5 per cent of the value of portfolio. This in turn implies an average taxable investment income of:

$4700 per month or $56 000 per year when the portfolio is at Objective #1; and
$5800 per month or $69 000 per year when the portfolio is at Objective #2

Once again, these estimates imply the existence of a significant income gap remaining at reaching both portfolio objectives.

Summary of results

So far historical data from the portfolio and three different approaches have been set out to seek to answer the question: how much income is the portfolio likely to produce?

Comparing estimates and income requirements

These individual estimates (blue) and the average of all estimates (green) are summarised in the charts below, and compared to the monthly income requirements (red) of both of the portfolio objectives. The chart below sets out the estimates for Objective #1. SAD Chart Ob1 The following chart sets out the same data and projections for the portfolio when it reaches Objective #2 (a portfolio total of $1 980 000). SAD Chart Ob2-corr The analysis shows that:

Portfolio income is likely to be below target at reaching Objective #1 – Using the approaches and history as a guide the portfolio should on average produce an income of around $57 000 per annum at Objective #1
And also below target at Objective #2 – When Objective #2 is reached portfolio income should on average be around $71 000
Therefore an income gap does exist to solve – Under most estimation approaches there will be a significant income shortfall at reaching both Objective #1 and #2
The gap is significant, but not disastrous – Assuming an equal weighting to the three approaches and actual historical distributions over the past three years the size of the income gap will be around $900 per month at Objective #1 (or $10 200 per annum) and greater, around $1000 per month at Objective #2 (or $12 000 per annum)
Only one estimation approach doesn’t identify a gap – Only if the ‘average distribution rate’ approach under Approach #1 is accurate will there be no income shortfall.

This implies that at the $1.6 million target of Objective #1, a small portion of any portfolio gains (around 0.6% of the value of the total portfolio) would need to be sold each year to meet this income gap. An identical result applies at the Objective #2, around 0.6% of value of the total portfolio would need to be sold annually.

Another intriguing implication of the reaching the average estimates is that it allows for an approximation of the required portfolio level to rely entirely on portfolio income, and avoid any sale of assets. At both portfolio Objectives the average of all estimation approaches indicate portfolio income of around 3.5 per cent.

Reversing this figure for the target portfolio income (e.g. for $67 000 at Objective #1 is 0.035/67000) implies a portfolio need of $1.91 million. For the higher target income for Objective #2, the implied portfolio required to not draw down capital is close to $2.4 million. This would require many additional years of future paid work to achieve.

Trailing clouds of vagueness

There are many caveats, inexactitudes and simplifications that should loom large in interpreting these results. The level of future returns as well as their income and capital components are unknowable and volatile.

In particular, the volatility of returns introduces key sequence of return risks that are simplified away by the reliance on deceptively stable historical estimates or averages. Particularly sharp movement in asset prices could change the asset allocation. Legislative or market changes could change the balance of income and capital appreciation targeted by Australian firms.

For these reasons, the analysis does not make me consider any particular remedial action. It indicates that under a range of assumptions and average outcomes, there will need to be a sale of some investments to meet the portfolio incomes targeted.

The same analysis shows that the superficially attractive choice to live only off portfolio income would in reality mean aiming for a target around 20 per cent higher – needing an extra $300 000 to $400 000 – potentially adding many years to the journey.

The relatively small scale of the required sales is the most surprising outcome of this analysis. Selling around 0.6 per cent of the portfolio annually does not on its face appear to be a high drawdown in most market conditions.

Another potential issue to consider is what this result means for asset allocation. There is no doubt that history would suggest that the income gap could be reduced by either reducing the bond allocation, or lower yielding international shares.

To give a sense of the magnitudes of this – using the ‘dead reckoning’ Approach #2 set out above – allocating 100 per cent of the equity portfolio to Australian shares would produce around $900 per month (or $10 300 per year) additional distributions at the Objective #1 portfolio of $1.6 million.

In theory, this domestic shares only option would all but close the income gap. Yet the benefits of diversification and risk reduction bonds and international shares offer make this a trade-off to consider, not a clear choice. At present, my plan would be to revisit this issue at my annual review of the portfolio asset allocation.

In the meantime, having produced these estimates has helped starting to think in more concrete terms about the draw down phase, its challenges and mechanics. In a small way, this seems to clear some of the clouds away, and enable me to glimpse some possible futures more clearly.

* Note: The historical average estimate for this purpose has been proportionally adjusted to increase based on the increased size of the portfolio between now and reaching Objective #2

The FI Explorer

Journeys in financial independence

Portfolio Analysis