My proposal for one fund with a smoothed return

[Colm Fagan]

I nearly wept this morning, hearing Regina Doherty on RTE with Rachel English, going over old ground of people having to decide whether they want to be low, medium or high risk; what happens if markets tumble on the day people retire; the carousel (probably one of the most ridiculous concepts ever devised); etc. I'm now absolutely convinced that the only way to avoid making a complete dog's dinner of AE is to adopt the smoothed, equity-based approach I've been advocating. A straightforward explanation can be found here. Additional papers and explanations can be found at http://www.colmfagan.ie/pensions.php

Unfortunately, it's proving difficult to get government, even my own profession, engaged. The prevailing view is that the proposals are too radical: "What you're proposing, Colm, hasn't been done anywhere else in the world; we don't want to be the first to try them out."

 

[Colm Fagan]

@Brendan Burgess
Hi Brendan
First of all, thanks again for organising the drinks the other night, and for supplying the goodies. If I'd known in advance I wouldn't have eaten my dinner beforehand!
In relation to AE, you could be right that government will succeed in setting up the scheme as proposed, with a panel of registered providers, a carousel default option, a range of risk-rated and specialist funds (for Muslims, ESG adherents, etc.), but it's going to be horrendously complicated and will most likely run into the sands like all previous efforts over the last 20 years and more. There's also a risk that it could turn into a mini version of the children's hospital cost fiasco. The plan as it stands is far too complicated.
In contrast, the approach I'm advocating is an administrative dawdle. It will be simple to implement a scheme where there's only one account for everyone, where there are no charges whatsoever other than a deduction from the credited interest rate, where everyone gets exactly the same rate of "interest", which is always positive (at least for the first 20 years) and which changes only once a quarter. You could nearly run it through the post office.
There can be a simple differentiation between AE and "standard" commercial pension schemes, which can continue in parallel. Tax relief (implied, since it will come as a government subsidy) will be at a fixed (say) 25% under AE, while a "standard" commercial pension arrangement will carry tax relief at the individual's highest marginal tax rate.
Your proposal for people to use some of their pension savings for a house deposit could be accommodated within the scheme I'm proposing, but I'm not a fan. I think it would just cause a rise in house prices if the resulting increase in demand is not accompanied by effective measures on the supply side.

 

[Brendan Burgess]

Hi Colm

Does this summarise it correctly?

Everyone’s pension savings will be invested in one fund.

That fund should be invested 100% in equities and property.

The underlying returns of such a fund would be very volatile – rising in some years and falling dramatically in other years.

However, the fund would smooth the returns so that the excess returns from the good years would be used to give positive returns in the bad years.

So everyone would earn a smoothed return on their investment which would not fluctuate wildly.

This return would always be positive - irrespective of the underlying investment performance.

Everyone would get the same return.

By investing in equities and property over the very long term, the returns will be far higher than a fund invested in bonds and cash.

Having one fund with one return for everyone will dramatically simplify the operation and the costs.

The fund will be a not-for-profit fund and will pay its costs out of the investment returns.

Higher returns and lower costs mean that people will have a much higher pot on retirement.

On retirement

The fund holder would get 25% of their fund tax free in the usual way.

The rest of the money would be left in the fund and would be drawn down over their expected life.

This would provide a far higher pension than an annuity and less volatility than an investment in an ARF.

A small part of each person’s fund would go into a longevity pot to make sure that their fund lasts their whole life if they live to a very old age.

 

[Brendan Burgess]

I think it's a great idea but could I make a suggestion to help get support for it.

Focus on the savings bit and not the retirement bit.

If you can convince people that on retirement, they will have a much bigger fund than they would otherwise have, that would be a huge achievement and you might get political buy-in for it.

If you can get this, then you can extend the idea to retirement.

If you succeeded in convincing people of the value of the scheme up to retirement but failed to convince them of the Lifetime Income Scheme, you would have made a huge contribution to the wealth and welfare of Irish citizens.

Brendan

 

[Colm Fagan]

Hi Brendan
Thanks. I really do appreciate your summary. A few comments/ elaborations:

This return would always be positive - irrespective of the underlying investment performance.

This is not an absolute guarantee, but is almost a certainty for at least the first 20 years. By then, everyone will fully understand how the scheme works, they will have years of strong positive returns under their belts and will be comfortable if there's the very occasional marginally negative quarterly return.

Focus on the savings bit and not the retirement bit.

I hear you, but a key advantage of the new approach is the high returns DURING retirement, because there's no de-risking. Even if the amount available AT retirement date is the same as under a more conventional arrangement, the pension from retirement under this approach will be MORE THAN DOUBLE that payable under an annuity, and the member doesn't have to cede control of their fund to an insurance company. See the attached slide, titled "Foot off the gas" to understand what I mean. Conventional pension arrangements involve taking your foot off the gas just when your fund is at its maximum earning power. Under my approach, the money stays invested in equities/property all through retirement.

 

[Brendan Burgess]

This is not an absolute guarantee, but is almost a certainty for at least the first 20 years. By then, everyone will fully understand how the scheme works, they will have years of strong positive returns under their belts and will be comfortable if there's the very occasional marginally negative quarterly return.

I don't understand why there would be an occasional marginally negative return?

The risk I see is from a sustained long-term fall in prices . A Japanese scenario. Within a few years of starting, prices fall over the following 10 to 20 years.

If your scheme starts in 2022 and underlying prices rise gradually over the next 10 years and then go into a long-term downturn, you could handle it.

Brendan

 

[Brendan Burgess]

but a key advantage of the new approach is the high returns DURING retirement, because there's no de-risking.

The problem is that are asking politicians and civil servants to do two dramatic and novel things which are not being done anywhere else in the World.

Start by trying to convince them of the first stage. And it's much easier to understand the concept of building up a fund to a retirement date.

The second stage is much more complicated - dealing with longevity, draw down rates,etc.

When we understand, accept and implement the first part, then unleash the second part on us.

My initial proposal on using the pension fund to buy a house faced many criticisms. I scaled it back to a very basic version "Allow the fund holder to borrow back his own money". That deals with a lot of the criticisms e.g. tax relief to buy houses and wall of money pushing up prices. If I can get this principle accepted and implemented, then we could review whether the next stage was worth going to.

Brendan

 

[Sarenco]

The risk I see is from a sustained long-term fall in prices . A Japanese scenario.

That's the key problem with Colm's proposal - it simply wouldn't survive contact with a particularly deep, prolonged draw down in equity/property prices.

 

[Colm Fagan]

Hi @Sarenco and @Brendan Burgess

Both your questions are good. Note however that they bring us into the honours class!

I don't understand why there would be an occasional marginally negative return?

That's the key problem with Colm's proposal - it simply wouldn't survive contact with a particularly deep, prolonged draw down in equity/property prices.

The formula to calculate the smoothed return is mechanical. There is no "actuarial discretion". Just twist the handle and take whatever comes out.
I did simulations based on historic returns going back to 1925 (US) and 1900 (UK). There was not one instance of a negative yearly return in any of those simulations.
With help from Brian Woods, I simulated 5,000 years of future returns (100 simulations, each for 50 years), assuming all sorts of mad movements in market values.
The earliest date for a negative yearly smoothed return (out of the 100 simulations) was in the fifth year. For this simulation, market values fell by 39% in the five quarters Q15 to Q19. The returns credited to accounts in the first five years were 6.6% (Year 1), 4.9% (Year 2), 6.1% (Year 3), 4.2% (Year 4) and -0.3% (Year 5). The return was back to +6.1% in year 6. All these figures are completely net of charges. I don't think anyone would complain about the -0.3% in year 5 in this scenario.
In papers to a Working Party of the Society of Actuaries in Ireland, I demonstrated that the approach WOULD survive a particularly deep, prolonged fall in equity and property values. I have put the various papers up on my website for anyone who wants to study them.

 

[Brendan Burgess]

Hi Colm

Let's say that between now and the start of 2022, there is a worldwide stockmarket boom.
Then we all start contributing to this fund. But the world stockmarkets crash slowly but surely over the next ten years.

If you "declare" a return of 5% a year when the fund has dropped in value by 5% a year, there will be a huge gap between the intrinsic value of the fund and the "declared" value of the fund.

Or am I missing something?

By the way, I think it's very unlikely. But it seems to me to be a possible outcome.

I accept that there could be a 50% crash in 2022. A level period in 2023 and then a stockmarket boom. The 2022 units might not recover, but the losses would be more than made up by the increases in the units purchased later.

But could there not be a slow but sure steady decline in stock prices? Maybe it has never happened.

Brendan

 

[Colm Fagan]

Hi Brendan. Yes, the simulations allow for all those possibilities. The "smoothed" value of the fund can - and does - depart significantly from the market value - in either direction. But market values are put on a pedestal they don't deserve. As you and I both know, they're irrelevant if you're not going to sell. Of course they're important in the long-run, but when there's no selling pressure you can adjust the carrying value of an asset to its market value over a very long period. That's what I'm doing.
Re-reading your post, I presume you mean "market value" when you write "intrinsic value". The "true value" or "intrinsic value" could differ significantly from either the market value or the smoothed value.

 

[Sarenco]

Fair enough Colm.

But grammar is important here. The UK/US simulations only relate to a relatively limited past, which isn't necessarily predictive of the future.

[As a side issue, can you remind me whether your simulations accounted for investment costs and taxes?]

And what about Japan? After all, it was the biggest stock market on the planet back in 1989.

 

[Colm Fagan]

@Sarenco I agree that the past is not a guide to the future. At the same time, a successful record over one hundred years plus of past experience shouldn't be sneezed at.
The 100 simulations of how the next 50 years might pan out were designed to address that weakness.
Arguably the worst of those simulations (discussed in the paper of 5 June 2019 to the Working Group of the Society of Actuaries - see my website) showed market values (net of the 0.5% management charge) falling more than 50% in 9 years and hadn't got back to the starting level after 13.5 years. Yet, even in that nightmare scenario, the smoothed approach survives. Needless to say, quoted returns aren't pretty, but they're a heck of a lot better than people would get in those circumstances from more traditional unit-linked products.

As a side issue, can you remind me whether your simulations accounted for investment costs and taxes?]

Yes, the figures are net of investment and admin charges (I assumed 0.5% for both combined, as per the DEASP's guidelines). It's a pension fund, so returns aren't subject to tax. (There's a problem of withholding taxes for some overseas assets, but that shouldn't be significant).

And what about Japan? After all, it was the biggest stock market on the planet back in 1989.

I didn't look specifically at Japan from 1989 for AE. I did look for the earlier paper on ARF's (which you can find here - from slide 38). The good thing about Japan from an AE perspective is that there was a massive run-up in Market Values in the period up to December 1989, so smoothed values would have been considerably less than market values at that time. This would provide a cushion. It's on my "to do" list to look at the numbers for Japan for AE, but I'm confident that it will come through OK. That's even before allowing for the fact that the fund will be well diversified in terms of geographies and asset classes (i.e. it will include real estate, infrastructure, possibly some private equity), so it won't be over-exposed to a single geography or industry sector (Japan had virtually no successful "modern tech" companies, a la Microsoft, Apple, etc.).

 

[Brendan Burgess]

I presume you mean "market value" when you write "intrinsic value".

Yes, I meant market value.

Brendan

 

[Duke of Marmalade]

Hi @Sarenco and @Brendan Burgess
The earliest date for a negative yearly smoothed return (out of the 100 simulations) was in the fifth year. For this simulation, market values fell by 39% in the five quarters Q15 to Q19. The returns credited to accounts in the first five years were 6.6% (Year 1), 4.9% (Year 2), 6.1% (Year 3), 4.2% (Year 4) and -0.3% (Year 5). The return was back to +6.1% in year 6. All these figures are completely net of charges. I don't think anyone would complain about the -0.3% in year 5 in this scenario.

I thought the 2007 crash was even worse than that and yet you say your formula survived that crash without negative returns

 

[NoRegretsCoyote]

The good thing about Japan from an AE perspective is that there was a massive run-up in Market Values in the period up to December 1989, so smoothed values would have been considerably less than market values at that time. This would provide a cushion.

But you have to pick a starting point!

You are right that returns will almost definitely be positive using any twenty-year moving average well into a hundred year period. But you can't start your fund in 2019 using the benefit of postitive returns since 1999. You have to start in 2019. Some people in the auto-enrolled system will only have the benefit of ten years paid-in contributions when they begin drawdown. Property and equities are volatile, and a bad run could leave them with little or even negative returns by then.

@Colm Fagan I think there is merit in your proposals - but I think this is one big flaw that's unaddressed.

 

[Colm Fagan]

I thought the 2007 crash was even worse than that and yet you say your formula survived that crash without negative returns

Hi Duke. The difference between the real world 2007 and the simulated Year 5 is that the historic 2007 smoothed value was below market value. This served as a cushion in the subsequent market fall. The same isn't true in the simulated Year 5 experience.

But you have to pick a starting point!

Yes, Coyote, you have to pick a starting point. It's the market value at the start. You don't speculate whether the smoothed value at the start should be lower or higher than the market value. I worried a bit about that too when I started down this road, but it's actually not that important in reality. In particular you don't "start your fund in 2019 using the benefit of positive returns since 1999". And yes, you do start in 2019. If you look at the formula, there is an inbuilt assumption that, in the long-term, equities will outperform bonds. The assumed outperformance in the formula is less than has been achieved in reality. You asked about people who will only have the benefit of 10 years' paid-in contributions when they begin drawdown. In that 10 years they will be given the benefit of a 3% (3.5% less 0.5% management charge) excess performance over bonds from day 1. That will be modified slightly as actual results unfold. In particular, if equities and property have a bad run, they won't suffer negative returns. Remember also that there is cross-generational solidarity, so if, for example, markets fall sharply in year 1 and recover from year 2 onwards, the people who joined in year 1 will share the benefits of the uplift in year 2 which, in a unit-linked fund, would be enjoyed exclusively by the year 2 joiners.
The simplest way to understand the points I'm trying to get across is to work through the example in Appendix 2 of my submission of 4 November 2018 to government (here). Appendix 2 is on pages 12 and 13. The example chosen to illustrate the formula in action is not dissimilar to the question you're asking. Market values fall by 2% in each of months 1 to 5, then rise by 4% in month 6. The monthly smoothed interest rate is 0.380% in month 1, then falls gradually in months 2 to 5, reaching 0.319% in month 5, then rises to 0.391% in month 6.
In summary, I have addressed what you think is a flaw. (I'm not saying the proposal is f lawless, though!!!).

 

[NoRegretsCoyote]

You asked about people who will only have the benefit of 10 years' paid-in contributions when they begin drawdown. In that 10 years they will be given the benefit of a 3% (3.5% less 0.5% management charge) excess performance over bonds from day 1. That will be modified slightly as actual results unfold. In particular, if equities and property have a bad run, they won't suffer negative returns.

Are you talking about some kind of guarantee? If so, who funds the guarantee?

 

[Colm Fagan]

Are you talking about some kind of guarantee? If so, who funds the guarantee?

It's simply a return that's calculated and added to the account. Yes, it eventually crystallises when the member cashes their fund (25% cashed at retirement, the balance over the remainder of their lifetime). So you can say that it is a guarantee by the time it crystallises. Who pays for the guarantee? It is paid for by the entire membership of the fund.
A simple example of an early crystallisation of the "guarantee" illustrates the point. Let's say that the fund falls 20% in month 1 and someone dies after paying a single contribution. That single contribution, plus a small amount of (positive) interest will be paid to their estate, even though the market value of that single contribution will have fallen 20%. Therefore, the fund (i.e. all the remaining contributors) has "lost" the difference between the market value of the deceased member's account and the amount that was paid to their estate. That loss reduces future returns for remaining members by a tiny amount in subsequent months/ years. I'm in a rush now, so may not have explained it well. I'll try to get back to you later if you're still uncertain.

 

[NoRegretsCoyote]

Who pays for the guarantee? It is paid for by the entire membership of the fund.

This is clear now.

You are proposing a pooling of risk across the contributors. The government's AE proposal is on the basis of individual accounts collectively managed, with no risk sharing.

This is neither good nor bad, but it is very different.